i

Generation, stability and migration of montmorillonite colloids in aqueous systems

Sandra García García

Doctoral Thesis

School of Chemical Science and Engineering

Royal Institute of Technology

Stockholm, Sweden 2010

AKADEMISK AVHANDLING

Som med tillstånd av Kungliga Tekniska Högskolan framlägges till offentlig granskning

för avläggande av Doktorsexamen i kemi fredagen den 29 januari 2010, kl. 10.00 i sal F1,

Kungliga Tekniska Högskolan, Lindstedtsvägen 22, KTH Stockholm

Opponent: Dr.Tiziana Missana

ii

ISBN: 978-91-7415-535-8

ISSN 1654-1081

TRITA-CHE Report 2010:3

© Sandra García García 2010

Tryck: Universitetsservice AB, Stockholm 2010

iii

To my family

iv

v

Abstract

In Sweden the encapsulated nuclear waste will be surrounded by compacted bentonite in

the granitic host rock. In contact with water-bearing fractures the bentonite barrier may

release montmorillonite colloids that may be further transported in groundwater. If large

amounts of material are eroded from the barrier, the buffer functionality can be

compromised. Furthermore, in the scenario of a leaking canister, strongly sorbing

radionuclides, can be transported by montmorillonite colloids towards the biosphere. This

thesis addresses the effects of groundwater chemistry on the generation, stability,

sorption and transport of montmorillonite colloids in water bearing rock fractures.

To be able to predict quantities of montmorillonite colloids released from the bentonite

barrier in contact with groundwater of varying salinity, generation and sedimentation test

were performed. The aim is first to gain understanding on the processes involved in

colloid generation from the bentonite barrier. Secondly it is to test if concentration

gradients of montmorillonite colloids outside the barrier determined by simple

sedimentation experiments are comparable to generation tests. Identical final

concentrations and colloid size distributions were achieved in both types of tests.

Colloid stability is strongly correlated to the groundwater chemistry. The impact of pH,

ionic strength and temperature was studied. Aggregation kinetics experiments revealed

that for colloid aggregation rate increased with increasing ionic strength. The aggregation

rate decreased with increasing pH. The temperature effect on montmorillonite colloid

stability is pH-dependent. At pH≤4, the rate constant for colloid aggregation increased

with increasing temperature, regardless of ionic strength. At pH≥10, the aggregation rate

constant decreased with increasing temperature. In the intermediate pH interval, the

aggregation rate constant decreased with increasing temperature except at the highest

ionic strength, where it increased. The relationship between the rate constant and the

ionic strength allowed the critical coagulation concentration (CCC) for Na- and Ca-

montmorillonite to be determined.

In order to distinguish the contribution of physical filtration and sorption to colloid

retention in transport, the different retention mechanisms were quantified. Sorption on

different representative minerals in granite fractures was measured for latex colloids (50,

100, 200 nm) and montmorillonite colloids as a function of ionic strength and pH.

Despite of the negative charge in mineral surfaces and colloids, sorption was detected.

The sorption is correlated to the mineral point of zero charge and the zeta potential of the

colloids, and increases with increasing ionic strength and decreasing pH. In transport

experiments with latex colloids in columns packed with fracture filling material, the

retention by sorption could clearly be seen. In particular at low flow rates, when the

contact time for colloids with the mineral surfaces were the longest, sorption contributed

to retention of the transport significantly. The retention of latex colloids appeared to be

irreversible in contrary to the reversible montmorillonite colloid retention.

Generation, stability and sorption of the montmorillonite colloids are controlled by

electrostatic forces; hence, the results were in qualitative agreement with DLVO.

vi

vii

Sammanfattning

I Sverige kommer kapseln som innehåller det urbrända kärnbränslet att omges med

kompakterad bentonitlera. Vi kontakt med vattenförande spricka kan bentonit avge

kolloidala partiklar som kan transporteras med grundvattnet. Om stora mängder material

eroderas bort från barriären äventyras dess funktion. I händelse av en läckande kapsel kan

lerkolloider bidra till spridning av starkt adsorberande radionuklider. I denna avhandling

studeras effekter av grundvattenkemi på generering, stabilitet, adsorption och transport av

montmorillonitkolloider i vattenbärande bergsprickor.

För att kunna förutsäga mängden monmorillonitkolloider som frisätts av

bentonitbarriären i kontakt med grundvatten av varierande salthalt har försök med

avseende om generering och sedimentation utförts. Dessa försök visade att den slutgiltiga

koncentrationen av kolloider är den samma i båda typerna av försök.

Kolloidstabilitet är starkt knutet till grundvattenkemi. Effekten av pH, jonstyrka och

temperatur har därför studerats. Dessa försök visade att hastigheten för

kolloidaggregering ökade med ökande jonstyrka och minskade med ökande pH.

Temperatureffekten är pH-beroende. Vid pH lägre än 4 ökar hastighetskonstanten för

aggregering med ökande temperatur, oberoende av jonstyrka. Vid pH högre än 10

minskar hastighetskonstanten med ökande temperatur. I pH intervallet 4-10 minskar

hastighetskonstanten med ökande temperatur utom vid höga jonstyrkor där trenden är den

motsatta. Sambandet mellan kinetik för aggregering och lösningens jonstyrka

möjliggjorde bestämning av den kritiska koaguleringskoncentrationen (CCC) för både

Na- och Ca-montmorillonit.

För att kunna skilja mellan retention pga fysikalisk filtrering och adsorption vid

kolloidtransport studerades dessa processer separat. Adsorption av latexkolloider (50, 100

och 200 nm) och montmorillonitkolloider på representativa mineralytor studerades som

funktion av jonstyrka och pH. Trots att både kolloider och mineralytor var negativt

laddade under rådande betingelser observerades adsorption. Affiniteten för kolloider till

mineralytor visade sig bero på nolladdningspunkten (pzc) för mineralet och

zetapotentialen för kolloiden. Affiniteten ökar med ökande jonstyrka och sjunkande pH. I

transportförsök med latexkolloider i kolonner packade med sprickfyllnadsmaterial kunde

retention pga sorption tydligt ses. Detta var tydligast vid låga flöden där uppehållstiden i

kolonnen är lång. Retentionen av latexkolloider var irreversibel i motsats till retentionen

av montmorillonitkolloider som visade tecken på reversibilitet.

Samtliga resultat är i god överensstämmelse med DLVO-teorin.

viii

ix

List of Publications

This doctoral thesis is based on the following papers and manuscripts, which are referred

to in the text by their Roman numerals:

I. Determining pseudo-equilibrium of montmorillonite colloids in generation

and sedimentation experiments as a function of ionic strength, cationic form,

and elevation

S. García-García, C. Degueldre, S. Wold, S. Frick

Journal of Colloid and Interface Science, 335 (2009) 54–61

II. Kinetic determination of Critical Coagulation Concentrations for Sodium-

and Calcium-Montmorillonite colloids in NaCl and CaCl2 aqueous solutions.

S. García-García, S. Wold, M. Jonsson

Journal of Colloid and Interface Science, 315 (2007) 512-519

III. Temperature effect on the stability of bentonite colloids in water.

S. García-García, M. Jonsson, S. Wold

Journal of Colloid and Interface Science, 298 (2006) 694-705

IV. Effect of pH and temperature on the stability of bentonite colloids.

S. García-García, S. Wold and M. Jonsson

Applied Clay Science, 43 (2009) 21-26

V. Colloid sorption on minerals. Effects of colloidal size, pH, buffer

concentration, and mineral properties

S. García-García, S. Wold and M. Jonsson

Submitted to Applied Geochemistry

VI. Colloid transport in fracture filling materials

S. García-García, S. Wold and M. Jonsson

Submitted to Journal of Contaminant Hydrology

Comment on my contribution to the publications

Paper I, I contributed to the design of the experiments and performed most of the

experiments. I have participated in the analysis of the results and I wrote the manuscript

in collaboration with the co-authors.

Paper II III IV V and VI, I have designed and performed the experiments. I have

carried on the simulations and participated in the analysis of the results and I wrote the

manuscript in collaboration with the co-authors.

x

xi

Table of Contents

Generation, stability and migration of montmorillonite colloids in aqueous systems ......... i 1. Introduction ..................................................................................................................... 1

1.1. Background of the project ........................................................................................ 1 1.2. Scope of the thesis ................................................................................................... 3

1.3. The bentonite barrier ................................................................................................ 4 1.4. The mineral montmorillonite ................................................................................... 6 1.5. Colloids .................................................................................................................... 7 1.6. Generation, aggregation and sedimentation of colloidal particles ........................... 7 1.7. The electrical double layer ....................................................................................... 9

1.8. Total inter-particle energy...................................................................................... 10 1.10. Colloid retention processes in water bearing fractures ........................................ 13

2. Materials ....................................................................................................................... 15 2.1. Bentonite ................................................................................................................ 15

2.2. Sodium- and calcium-montmorillonite .................................................................. 16 2.3. Reference colloids .................................................................................................. 17

3. Methods......................................................................................................................... 21 3.1. Single Particle Counting ........................................................................................ 21 3.2. Photon Correlation Spectroscopy .......................................................................... 21

3.3. Zeta potential analysis............................................................................................ 23 3.4. Fluorescence spectrophotometry ........................................................................... 23

4. Results and Discussion ................................................................................................. 25 4.1. Colloid generation .................................................................................................. 25

4.2. Stability of colloidal suspensions........................................................................... 29 4.2.1. Method applicability and PCS calibration .................................................. 29

4.2.2. Effect of the ionic strength on aggregation kinetics ................................... 31 4.2.3. Influence of temperature on aggregation kinetics ....................................... 37 4.2.4. Ionic strength, pH and temperature effects on the stability of

montmorillonite colloids ....................................................................................... 41

4.3. Adsorption of colloids on mineral surfaces ........................................................... 45 4.4. Colloid transport .................................................................................................... 47

5. Concluding remarks ...................................................................................................... 53 6. Acknowledgements ....................................................................................................... 54 7. Bibliography ................................................................................................................. 55

xii

1

1. Introduction

1.1. Background of the project

Colloidal particles suspended in deep bedrock groundwaters are generated as the result of

a number of processes such as mineral weathering, precipitation, erosion, biological

activity and organic matter degradation. The suspended particles have diameters ranging

from nanometers to micrometers, and varying chemical compositions.1-4

The small size

of colloidal particles gives a large surface-to-volume ratio as well as lesser effect of the

gravitational force. In addition, the surface groups of colloids often give raise to surface

charge in natural waters. These properties allow colloidal particles to remain in

suspension, to diffuse by Brownian motion and to sorb contaminants from the

groundwater by electrostatic interactions, surface complexation, precipitation,

polymerization or ion exchange.5 Association of contaminants with mobile colloidal

particles, is an important environmental issue since contaminants are often spread with

colloids in nature.6,7

Hence, understanding the mechanisms of generation, stability and

mobility of colloidal particles in nature is of great importance for predicting the fate of

contaminants in the environment.

One particular case where colloid formation and migration is of crucial importance is

deep geological repositories for highly radioactive nuclear waste. Deep geological

repositories for spent nuclear fuel are constructed to isolate the radioactive waste from

the biosphere for a period of time as long as 100.000 years. Many countries carry out

active research programs in order to take care of their own nuclear waste. The main

repository alternatives are natural clay and crystalline rock formations. Clay formations

have been adopted by France and Belgium,8,9

while Finland and Sweden will build their

repositories in crystalline rock.10,11

The company Posiva Oy in Finland has already

started to excavate the tunnel at the Olkiluoto site, and The Swedish Nuclear Fuel and

Waste Management Co (SKB) in Sweden has recently suggested Forsmark as the most

suitable location for the repository. The disposal facilities in Sweden and Finland will

both be constructed in granite bedrock in accordance with the KBS-3 concept developed

by SKB.

The Swedish KBS-3 concept will fulfill the safety requirements by a multi-barrier system

consisting of four engineered and natural barriers12

as represented in Figure 1. In a multi-

barrier system, the different barriers support and complement each other but they are

intended to be functionally independent in the event of the possible failure of any of the

other barriers:13,14

- The spent fuel itself constitutes the first barrier. The spent fuel consists of 95% UO2 and

5% fission products and actinides. In the event of groundwater intrusion due to barrier

failure, radionuclides will be released at the rate of UO2 (fuel matrix) dissolution.

Uranium dioxide has low solubility under reducing conditions.15

2

- The second barrier is a copper canister provided with a cast iron insert in which the fuel

elements will be encapsulated. The copper canister is does not corrode under reducing

conditions.16

- The third barrier is the compacted saturated bentonite that will surround the canisters. It

will provide mechanical support for the canisters. Another beneficial feature is that the

plasticity of the swollen bentonite will minimize the potential damage caused to the

canister by any future rock movements.17

When water-saturated, the bentonite barrier has

low hydraulic permeability which will reduce water flow around the canisters. Due to the

low hydraulic permeability, the only transport mechanism through the barrier is

diffusion,18

which will limit the access of corrosive species such as HS- to the canister

and will retard escaping radionuclides. The migration of positively charged radionuclides

will be further retarded by surface complexation and cation exchange processes in the

bentonite matrix.

Figure 1: The KBS-3 concept for disposal of spent nuclear fuel.6

- The crystalline granite host rock is the fourth barrier that ensures a stable mechanical

and chemical environment to the canisters and where sorbing surfaces for radiunuclides.19

The host rock has to fulfill a number of requirements in order to be suitable for a deep

repository. One important safety requirement is to have a low fracture frequency since the

presence of groundwater is undesirable due to the risk of canister corrosion and transport

of radionuclides in groundwater.20

Of the two sites that were under consideration in

Sweden, this requirement is best fulfilled by the Forsmark site. Groundwater at the

Forsmark site is generally anaerobic with reported 0.032-0.052 mL L-1

dissolved oxygen.

It has a pH of 7.7 and a redox potential (Ag/AgCl) of –281 mV. The natural colloid

content measured by Laser-Induced Breakdown Detection (LIBD) was 30–50 μg L-1

at

3

437 m depth.21

The flow measured in 32 boreholes was found to be 0.01 to 123 mL min-1

with calculated Darcy velocities from 6.7∙10–11

to 6.6∙10–7

m s-1

.22

Although the presence of fractures will be avoided, there will be some water-bearing

fractures that will intercept the deposition holes. Under present conditions in the selected

host rock, release and transport of colloidal particles is not likely to occur since the

groundwater salinity is high. However the physicochemical conditions in permeable

fracture formations may change23,24

and thereby enhance or inhibit colloid transport.

For instance, in contact with low saline groundwaters after calcium depletion,25,26

a

highly hydrated bentonite gel may form, consisting of charged aluminosilicate sheets

weakly bound to each other by electrostactic and van der Waals forces. At the bentonite-

water interface, disintegration of the gel into colloidal particles could take place if

repulsion between charged sheets is strong enough. If bentonite colloidal particles are

released from the buffer barrier in large quantities, the functionality of the bentonite

barrier will decrease. Loss of bentonite material would imply higher hydraulic

conductivity, which in turn will allow corrosive species to reach the canisters faster,

thereby increasing the risk of corrosion and radionuclide leakage.

1.2. Scope of the thesis

The aim of this work is to understand montmorillonite colloid stability and transport in

water-bearing fractures. More specifically, this has been done by:

- Studying colloid generation from Na- and Ca-montmorillonite as a function of ionic

strength.

- Studying the effect of water chemistry on colloid stability; the impact of ionic strength,

pH and temperature on Na- and Ca-montmorillonite.

- Studying the sorption of colloids on mineral surfaces at static conditions.

- Studying the migration of montmorillonite colloids in columns packed with fracture

filling material; the role of retention processes.

A schematic illustration of the scope of this thesis is given in Figure 2.

4

Bentonite barrier

in contact with

Water-bearing fractures

1. Does Colloid generation occur?

(Paper I)

2. Are Colloids stable?

(Paper II, III, IV)

3. Do Colloids sorb onto Host Rock?

(Paper V)

4. What is the impact of retention

on Colloid transport?

(Paper VI)

DLVO

Theory

Figure 2: Schematic representation of the scope of this thesis.

1.3. The bentonite barrier

The motivation for using bentonite as a buffer material is its high content (65-90%) of the

mineral montmorillonite. Montmorillonite has pH buffering capacity and swells in

contact with water. These properties will provide a stable physical and chemical

environment for the canisters (see Figure 3). When saturated with de-ionized water to a

density of 2000±50 kg m-3

, compacted bentonite has a swelling pressure of about 104 kPa

and a hydraulic conductivity27

of 7∙10-14

m s-1

. Therefore, the water flow around the

canister will be reduced by the buffer barrier, preventing corrosive agents such as

sulphide from coming into contact with the canister.25

In addition, montmorillonite has excellent sorption characteristics. In the event of

radionuclides being released from a failed canister, the mobility of cationic contaminants

will be reduced, since they will be retarded by surface complexation and cation exchange

reactions.

During the lifetime of the repository, the fracture system, the groundwater composition

and flow may be affected by climate changes. The potential changes in the climate and

the consequences for the repository are difficult to predict.

Scandinavia is expected to experience glacial cycles, during which the ice sheet grow and

retreat and melt water accumulates and can infiltrate by pore pressure differences

generated at both sides of the ice sheet margin. If larger fractures open by the pressure

5

released after ice retreat, the melt water may with high flow rates, reach the repository,

displacing the more saline pre-existing groundwaters without mixing.23,24

In this scenario, these processes will lead to hydrological changes in permeability,

groundwater pressure, groundwater flow, groundwater salinity, pH and oxygen content at

repository depths.28

Figure 3: Deposition chamber with bentonite buffer and canister.

14

The importance of colloid release lies not only in the loss of buffer material, but also in

the risk of colloid-facilitated radionuclide transport, since in the event of canister failure

most radionuclides will adsorb strongly onto bentonite. It is known that the migration

velocity of soluble contaminants in groundwater is retarded due to matrix sorption.29

However, when sorbed to colloids, contaminants get mobilized and migrate faster.7,30

6

1.4. The mineral montmorillonite

Montmorillonite is a layered aluminosilicate mineral from the 2:1 smectite group. Each

layer consists of three sheets: an octahedral sheet between two tetrahedral sheets, as

illustrated in Figure 4. Aluminium atoms are present at the octahedral sites, coordinated

to oxygen and hydroxyl groups, while silicon atoms coordinate to oxygen in the

tetrahedral positions.

Figure 4: Na-montmorillonite structure.

Isomorphous substitution occurs mainly in octahedral sheets, where Al is replaced by Mg

or Fe, but substitution of Si by Al in the tetrahedral sheets can also take place. As a

consequence, the layers have a permanent negative charge with a charge density of about

0.13 C m-2

. The excess of negative charge is compensated for by adsorbed positive ions

such as Na+ and Ca

2+ between the layers. The binding forces between layers are much

weaker than those within sheets. When montmorillonite comes into contact with water,

exchangeable cations hydrate and successive water layers occupy the interlayer space and

the montmorillonite swells. The extent of swelling depends on the compensating cations

and available volume. Under constricted volume, the incorporation of well ordered water

layers is known as crystalline swelling. Under free-swelling conditions, more than four

water layers may be incorporated by Na-montmorillonite by osmotic swelling until

repulsive and attractive forces between sheets reach equilibrium. Ca-montmorillonite on

the other hand, shows little if any osmotic swelling.31

Once the montmorillonite layers

have been filled with water molecules, the compensating cations can easily diffuse out

and be exchanged by new ions from the solution/liquid phase. This feature is

characteristic of many clay minerals. The typical cation exchange capacity in

montmorillonite is about 0.9 meq g1.18

7

1.5. Colloids

A colloidal dispersion is a distribution of small particles of a substance (solid, liquid or

gas) in a continuous phase. The size of the particles ranges from nanometers up to

micrometers.

Given the small size (1 nm- 1 µm) of the suspended particles, they have a large surface

area per unit mass. Due to their low mass, colloidal particles are not strongly affected by

gravitational force. Most particles in suspension bear a surface charge, which results in

repulsive forces between particles, preventing them from agglomerating. Voluminous

groups binding or adsorbing on the surface can also prevent particle agglomeration by

osmotic and volume restriction effects.32

The combination of a low tendency to

agglomerate and the low effect of the gravitational force results in stable colloidal

dispersions.

Similar to ions and molecules, colloidal particles in a fluid undergo random

displacements and collisions. This phenomenon is known as Brownian motion. As a

result, colloidal particles are transported by diffusion and distribute homogeneously in a

liquid. 33

Depending on the water affinity of the dry solid, colloids are traditionally classified as

lyophilic or lyophobic, where the lyophobic colloids have lower affinity for water.

Another common classification used refers to the stabilization mechanism that prevents

particles from sticking together, i.e., steric stabilization or electrostatic stabilization.34

Depending on the size distribution, colloids can be classified as monodisperse or

polydisperse. In principle, in a monodisperse suspension all the particles have the same

shape and size, while in polydisperse suspensions the particles differ in size and shape.34

Most of the organic and inorganic colloids in natural systems have broad size

distributions.35,36,37

1.6. Generation, aggregation and sedimentation of colloidal particles

Particles of clay minerals, iron hydroxides, silica and natural organic matter among others

are present in natural aquifers in the colloidal size range and their concentrations are

affected by hydrogeochemical perturbations.38,39

Montmorillonite colloid generation can occur when the material is in contact with water.

Swelling leads to the formation of a gel-like front moving with the montmorillonite

concentration gradient. Due to high hydration of montmorillonite in the gel structure, the

attractive forces between the negative sheets and the hydrated cations become weaker and

colloidal particles start to disjoin and diffuse from the montmorillonite/water interface.

This is the mechanism for generation of montmorillonite colloidal particles. In the case of

8

Ca-montmorillonite the swelling capacity that induces gel propagation is very low

compared to Na-montmorillonite.

The swelling pressure of montmorillonite, and hence colloid generation, increases with

decreasing ionic strength, increasing compaction density and pH.40

This can be

understood by considering the effects of these parameters on the electrostatics on the

colloidal montmorillonite particles.

When two particles collide they can form an aggregate. If the particles undergo

irreversible aggregation, the colloids coagulate, while if the aggregates may be

redispersed by shaking or changing the conditions in the system, the colloids flocculate.

Flocculated lyophobic suspensions redisperse if the conditions change leading to an

increase in the stability of the particles.

In order for small particles to sediment, it is necessary to apply a centrifugal field

stronger than the normal gravitational field. Large aggregates deposit at a uniform rate

determined by the ratio between the gravitational and friction forces. The sedimentation

rate of a large aggregate due to the gravitational field can be expressed as:

fgmfgvdtdx /)/1(/)(/ 00 (1)

where g is the acceleration due to gravity in N kg−1

, m is the mass of the particle in kg, v

is the volume of the particle (m3), ρ is the density of the particle (kg m

-3), ρ0 is the density

of the solution and f is the frictional coefficient, proportional to the viscosity of the

medium η in Pa s and the radius of the particles a in metres.

41

af 6 (2)

After aggregation the probability of sedimentation increases, since both processes are

intimately connected. When aggregation is the limiting step, sedimentation of the large

aggregates is faster than with their rate of formation. Hence, in this case the size

distribution of the particles does not change significantly with time in the upper part of

the suspensions and aggregation exhibits second order kinetics (Paper II, III, IV).

Colloid aggregation can be represented by a bimolecular reaction:42

A + A → A2 (3)

For reaction (3), the rate can be expressed as:

22 Akdt

Ad (4)

where k is the rate constant for the aggregation process and [A] is the concentration of

particles. Integration of equation (4) leads to:

9

ktAA t 2/1/1 0 (5)

The slope of the line obtained when 1/[A]is plotted versus t is 2k.

The aggregation rate for particles in the diffusion-controlled regime is given by

expression (6):43

216/ AaDdtAd p (6)

where t is time, Dp is the diffusion coefficient and a is the particle radius. The diffusion

coefficient, Dp can be defined as:

aTkD Bp 6/ (7)

where kB is the Boltzmann constant, T is temperature and η is the viscosity of the

medium. Expression (6) can be rewritten as:

23

8/ A

TkdtAd B

(8)

where 3

8 TkB is the diffusion-controlled rate constant, which is independent of the radius

and nature of the particles and only depends on the viscosity of the medium. For dilute

aqueous suspensions, the diffusion-controlled rate constant for colloidal particles is

6.53∙109 l mol

-1 s

-1.

1.7. The electrical double layer

In electrostatically stabilized suspensions, the surface charge of a colloidal particle

attracts ions of the opposite charge (counter-ions) that attach firmly, building the so-

called Stern layer. More counter-ions are then attracted in order to neutralize the charged

particle but these repel each other and are repelled by the ions in the Stern layer.

Therefore, they form a dynamic diffuse layer of counter-ions. The concentration of

counter-ions in the diffuse layer gradually decreases with distance from the surface, until

it reaches the concentration in the bulk. The Stern layer and diffuse layer constitute the

electrical double layer. The thickness of the electrical double layer depends on the type

and concentration of the ions in the suspension, the particle surface, temperature, etc. The

Debye-Hückel parameter κ is the inverse thickness of the electrical double layer defined

as:

2/1

0

0

22

Tk

nze

Br

iii

(9)

10

where e is the elementary charge, z is the ion charge, n0 is the number of ions per cubic

metre, εo is the permittivity of vacuum, εr is the dielectric constant, kB is the Boltzmann

constant and T is the absolute temperature.

1.8. Total inter-particle energy

The DLVO theory, named after Derjaguin, Landau, Verwey and Overbeek, describes the

interaction of two charged particles as the total energy (VT) that results from the sum of

the electro-osmotic repulsion between the ionic clouds and the Van der Waals attraction:

ART VVV (10)

The repulsive energy (VR) of the particles as a result of the interactions between their

diffuse-double layers can for spherical geometry be approximated by the expression:

kxB

R eTkan

V 22

2

064

(11)

where a is the particle radius, n0 is the number of ions per unit volume in the bulk, x is the

distance of interaction, κ is given by expression (9), and γ is a factor relating to the

surface potential (ψo) through the less restricted Gouy-Chapman expression (12), where z

is the valence of the electrolyte:

12

exp

12

exp

0

0

Tkze

Tkze

B

B

(12)

The attractive Van der Waals energy (VA) for spherical particles is given by the

expression:

2

2

2

2

2

2

2

4ln

2

2

4

2

6 ax

axs

ax

a

axx

aHVA (13)

where H is the Hamaker constant defined as:

2/32

2

2

1

22

2

2

1

2

21

21

216

3

4

3

nn

nnhTkH eB

(14)

where ε1 and ε2 are the dielectric constants for the material and the medium, h is Planck´s

constant, ve is the mean electronic adsorption frequency in the UV spectrum of the

11

medium and n1 and n2 are the refractive index in the visible spectrum for the material and

the medium, respectively.44,45

The total energy function VT displays a maximum at a certain particle distance. The

maximum of the total energy function corresponds to the energy that the particles must

surmount to aggregate. When the maximum of the total energy is negative, attraction

between particles will dominate and the system will be unstable, since every collision

between particles will lead to aggregation. On the contrary, when the total energy

maximum is positive, only the collisions with enough energy to overcome the energy

barrier will form an aggregate. Upon increasing the energy barrier, the fraction of

collisions with sufficient energy to overcome the energy barrier will be reduced, resulting

in a more stable suspension.

The height of the energy barrier depends on the surface potential and the electrolyte

concentration. Increasing electrolyte concentration reduces the double layer thickness, as

can be deduced from equation (9). The double layer compression reduces the repulsive

energy between particles according to equation (11). The electrolyte concentration at

which the repulsive energy is equal to the attractive energy is called the critical

coagulation concentration, CCC.46

At CCC and higher electrolyte concentrations,

colloidal suspensions are unstable. The height of the maximum total energy, which can

be compared with the activation energy for particle aggregation, is zero. Therefore every

collision between particles forms an aggregate, and the aggregation process takes place in

the diffusion-controlled regime.

1.9. Diffusive, advective (or conductive) and dispersive transport

The mechanisms for colloid transport through water-saturated geologic medium include

advection due to bulk motion of the fluid (also called convection) and dispersive

transport. Dispersive transport is caused by diffusion (similar to molecules) and

mechanical mixing by velocity variations in the porous matrix. Mechanical mixing

includes differences in fluid velocity in the center of the pores compared to the edges,

differences in pathway lengths and differences in velocity in larger and smaller pores.

Diffusion is the dominating transport mechanism at groundwater velocities lower than 1.6

10-10

m s-1

.47,48

The diffusion of colloidal particles in groundwater is described by the

Fick’s first law. The Fick´s first law establish the relationship between the mass flux per

unit time per unit area FDiff (kg m-2

s-1

) and the concentration gradient dC/dx, (kg m

-3)

where s is the distance in m and D is the effective diffusion coefficient of the medium

with units of m2·s

-1:

dx

dCDFDiff (15)

12

The negative sign in the above equation indicates that particle migration is in the

direction of decreasing concentration. The Fick’s first law is modified for a saturated

porous media according to the following expression:

dx

dCDF eDiff (16)

where φe is the effective porosity of the medium.

Total porosity (φ) is the ratio of volume of void space or interstices (Vv) to the total

volume (Vt).49

t

v

V

V (17)

Effective porosity (φe) is the volume that corresponds to interconnected pores through

which water can flow. Thus, not all the void space contributes to groundwater flow and

effective porosity is lower than total porosity since it does not include non-interconnected

pores.

Advective transport (or convective transport) occurs when colloidal particles move with

the water front. The flux for advection FAdv is expressed as:

CqCnvF xexAdv (18)

where FAdv represents the mass of colloid per unit cross-sectional area transported in the x

direction per unit time, qx is the Darcy velocity in the same direction,50

ne the effective

porosity, and vx the groundwater velocity or pore water velocity in the same direction.

Transported colloidal particles are subject to longitudinal or mechanical dispersion due to

velocity variations and differences in travel time along different flow paths. The

mechanical dispersive flux (or dispersive flux) FDisp can be described similarly to the

diffusive flux:

dx

dCDF emDisp (19)

where Dm is the mechanical dispersion coefficient.

The total flux FT in porous media will be the sum of the above introduced three flux FDiff,

FAdv and FDisp:

DispAdvDiffT FFFF (20)

Substituting equations (16), (18) and (19) on (20) we obtain:

13

dx

dCDCvF LexeT (21)

where DL is the longitudinal diffusion coefficient which corresponds to the sum of the

effective molecular diffusion coefficient D, and the Dm is the mechanical dispersion

coefficient:51

mL DDD (22)

The advective–dispersive (or convective-dispersive) transport is to some extent a

probabilistic process, since which flow line particles will step over at a grain boundary

follows statistics. Therefore, the concentration of particles breaking through after

transport in a porous medium describes a breakthrough curve rather than a narrow peak.52

1.10. Colloid retention processes in water bearing fractures

Particle sedimentation, attachment to surfaces and filtration are the main retention

mechanisms that prevent colloid transport. Figure 5 summarizes all the processes relevant

to colloid transport in natural fractures.

Figure 5: Colloid transport in natural fractures.

14

Suspended colloidal particles in groundwater flowing through a porous medium such as

fracture filling materials under advection regime might interact by sorption with the

mineral surfaces, become entrapped in pore constrictions and surface roughness

(filtration), disperse in the porous matrix (mechanical dispersion) or continue flowing

without capture.

Sorption is the general term used when the retention mechanism of ions or molecules

(sorbate) at a solid surface (sorbent) is unknown. Sorption includes adsorption, surface

precipitation and polymerization processes.53

The solid-water distribution coefficients (Kd) are widely used by geochemists to

determine the accumulation of solute (sorbate) at solid surfaces (sorbent) in an aqueous

medium. It should be noted that the determined Kd values do not provide information

about sorption mechanisms, that can be either by adsorption, surface precipitation or

polymerization, involving both physical and chemical processes.

For sorption processes independent of the concentration of sorbate, the distribution

coefficient, Kd, is a constant that expresses the ratio of equilibrium colloid concentrations

sorbed and in solution, according to the following expression:

m

V

C

CCK

f

fi

d

(23)

where Ci and Cf are the initial and final sorbate concentrations in solution, V is the

volume of solution in m3 and m is the mass of sorbent in kilograms.

By normalizing the Kd values to the specific surface area sa of the sorbent, comparison of

sorption between different sorbents becomes more straight forward.

af

fi

dsm

V

C

CCK

(24)

where sa is the surface area of the mineral in m2 kg

-1 usually determined by BET

measurements. Thus, the surface normalized Kd values have units of m.

15

2. Materials

2.1. Bentonite

Bentonites of different origin differ in exchangeable cations, montmorillonite content,

number and abundance of accessory minerals. The Wyoming form of bentonite (MX-80)

is widely used as reference material to investigate different properties of the buffer and

backfilling material in the deep repository. The bentonite used in this study is Wyoming

MX-80 supplied by the American Colloid Co.

The mineral composition of MX-80 is given in Table 1.

Table 1. Mineral composition of bentonite Wyoming MX-80.

18

Component MX-80

(wt-%)

Uncertainty

(± wt-%)

Calcite + Siderite 0–1 1

Quartz 3 0.5

Cristobalite 2 0.5

Pyrite 0.07 0.05

Mica 4 1

Gypsum 0.7 0.2

Albite 3 1

Dolomite 0 1

Montmorillonite 87 3

Na- 72% 5

Ca- 18% 5

Mg- 8% 5

K- 2% 1

Anorthoclase 0 1

CEC (meq 100 g-1

) 75 2

Organic carbon 0.2 –

Two different procedures were used to prepare colloidal bentonite suspensions.

For one series of experiments, (Paper III), suspensions were prepared from commercial

bentonite Wyoming (MX-80) as received from the supplier. One gram of bentonite was

dispersed in 250 ml of 10-3

M NaCl or NaClO4 solutions and the suspension was allowed

to settle for 24 h. Then 80 ml of the upper colloidal part were collected and used in the

experiments. The colloid concentration was determined gravimetrically to be 0.88 0.06

mg ml-1

. The pH of the suspensions was measured to approximately 9.

16

In other series of experiments (Paper IV), soluble and coarse minerals were removed

from bentonite as follows: (i) 0.3 g of clay was dispersed in 30 ml de-ionized water and

centrifuged for two hours at 6000 rpm. (ii) The supernatant solution was decanted off.

(iii) The sediment was collected and the clay fraction was separated from the coarser

minerals. (iv) The clay fraction was dispersed in 30 ml de-ionized water and centrifuged

at 6000 rpm for two hours. The conductance of the rejected supernatant was <80 μS. (v)

The sediment was collected and oven-dried at 60°C.

A stock colloid suspension of 0.8 g l-1

was prepared from the purified material. The initial

pH of the suspension was 8.3±0.1. Small volumes of NaOH was used to adjust the pH to

10 and 11. NaOH additions were buffered by bentonite and the pH of the suspensions

returned to the initial value. Daily additions of NaOH were necessary for at least two

weeks until the pH of the suspensions was stable. Once the basic pH was stable, the

acidic pH was adjusted to 3, 4 and 6 by adding small HCl volumes. These values were

achieved immediately after adding HCl and were constant so no further additions were

necessary. By conductance measurements, the concentration of NaOH was determined in

the suspensions with pH 10 and 11. The corresponding amount of NaCl was added to the

suspensions with pH 3, 4, 6 and 8.3 in order to obtain equal total ionic strengths. The

total ionic strengths studied were 0.002, 0.003, 0.005 and 0.0068 M. The pH of the

suspension with ionic strength 0.005 M was adjusted to 10 by using the buffer pair

NaHCO3/Na2CO3 instead of NaOH for comparison.

2.2. Sodium- and calcium-montmorillonite

The sodium- and calcium-montmorillonites used (Paper II, V, VI) were obtained by

purifying Wyoming bentonite MX-80 at the laboratory of Clay Technology AB. The

purification procedure was as follows: A 10 g portion of MX-80 bentonite was dispersed

in 1 L of 1 M analytical grade chloride solution of the desired cation and left to settle.

The supernatant was removed and the procedure repeated three times. The material was

washed three times with de-ionized water and the supernatant was removed after

centrifugation. The clay fraction suspension was separated from the accessory minerals

by decanting. In order to remove excess electrolytes, the clay suspension was transferred

to dialysis membranes (Spectrapore 3, 3500 MWCO) placed in plastic containers with

de-ionized water. The water was changed daily until the electrical conductivity was

below 10 μS cm-1

. The material was then redispersed in 1 L of 1 M analytical grade

chloride, and the process was repeated again. The montmorillonite was then dried at 60oC

and milled to an aggregate grain size similar to that of MX-80. Thereafter, the structural

formula of the Na-exchanged montmorillonite was determined by ICP-AES element

analysis as:18

(Na 0.64 K 0.01)(Al 3.11 Ti 0.01 Fe 3+

0.36 Mg 0.47)[( Si 7.93 Al 0.07)O20](OH)4 x nH2O (25)

and that of the Ca-exchanged form as:

17

(Ca 0.25 Na 0.01 K 0.01)( Al 3.14 Ti 0.01 Fe 3+

0.37 Mg 0.47 )[(Si 7.93 Al 0.04) O20](OH)4 x nH2O

(26)

Purified sodium- and calcium-montmorillonite were used to prepare suspensions of 0.8 g

L-1

.The dry material was poured into deionised water and the suspension was put into an

ultrasonic bath for two hours (Paper II). The pH measured in the suspensions was 8.3 ±

0.1 The ionic strength was adjusted by addition of small volumes of concentrated NaCl

and CaCl2 solutions.

In sorption experiments (Paper V), 30 mL of Na-montmorillonite colloidal suspensions

with a concentration of 5 x 10-4

g L-1

were in contact with different amounts of quartz

sand using buffered solutions 0.001 M of TRIS (tris(hydroxymethyl)aminomethane) at

pH 8.5 and MES (2-(N-morpholino)ethane-sulfonic acid) at pH 6 and 4.5.

For transport experiments performed in Paper VI, 0.05 g of dry clay was treated

ultrasonically until complete dispersion in 2 mL deionized water containing the

conservative (non-sorbing) tracer tritium (HTO) 600 Bq cm-3

(TRY64, 2 MBq, 5 g,

Canberra).

2.3. Reference colloids

Carboxyl and amidine terminated surfactant-free polystyrene latex particles purchased

from Interfacial Dynamics Corporation (IDC, Portland, USA) were used as reference

colloids in stability tests at different temperatures (Paper III). Suspensions were prepared

by diluting the stock supplied to particle concentrations in the range 0.38-3.10 mg mL-1

in

10-3

M NaCl or NaClO4 solutions.

For sorption and transport experiments (Papers V VI), fluorescent polystyrene latex

colloidal particles (Duke Scientific Corporation, CA, USA) were used. Commercial

names G50B, R100B and B200B correspond to 50±5, 100±5 and 200±10 nm diameter

respectively. Batch sorption experiments were performed in 0.001 and 0.01 M buffer

solutions, using TRIS at pH 8.5 and MES at pH 6 and 4.5.

Table 2 provides size, zeta potentials of reference and natural colloids.

The concentration of fluorescent particles in batch sorption and in colloid transport

experiments was (5±0.2) x 109 particles mL

-1 for 50 nm, (1.0±0.2) x 10

10 particles mL

-1

for 100 and 200 nm respectively.

The latex colloidal cocktail was introduced in the column as a 1 mL single injection

containing and tritium (HTO) 600 Bq cm-3

used as conservative tracer.

18

Table 2. Diameter and zeta potential of the colloidal particles at 0.001 M.

Diameter

(nm)

Temperature

(ºC)

Zeta potential (mV)

pH=8-8.5 pH=6 pH=4.5

Bentonite 20-600 0-20-80 -508a

Na-mont.+NaCl 20-600 20 -54±9a

Ca-mont.+NaCl 20-600 20 -47±8a

Na-mont.+CaCl2 20-600 20 -25±5a

Ca-mont.+CaCl2 20-600 20 -19±4a

Amidine Latex 490±11 20 +7410a

Amidine Latex 490±11 80 +74a

Fluorescent Latex 50±5 22 -33±1b -31±1

b -22±1

b

Fluorescent Latex 100±5 22 -45±2b -45±2

b -45±2

b

Fluorescent Latex 200±10 22 -50±2b -50±2

b -42±2

b

Na-montmorillonite 20-600 22 -38±2b -41±2

b -41±2

b

a Measured by means of a ZetaPALS Zeta Potential Analyzer, Brookhaven Instruments Corporation

b Measured by means of a Malvern Zetasizer 3, Malvern Instruments.

2.4. Minerals

The minerals used in Paper V and VI were quartz, granite and the fracture filling

materials epidote, biotite and calcite.

Quartz sand was purchased from Fluka, 99.999% SiO2. Quartz with BCR reference

material No.131, and a size range of 0.48-1.8 mm, was used in sorption experiments

(Paper V) and quartz BCR reference materials No.068 and No. 132, with size ranges of

0.16-0.63 and 1.4-5 mm respectively was used in transport experiments (Paper VI).

Granite, epidote and calcite were sampled at the Äspö Hardrock Laboratory in Sweden

and grinded. The 0.5-2 mm fraction was used in sorption experiments.

Characterization of granite, epidote and calcite obtained at Äspö Hardrock Laboratory

was carried out by powder X-ray diffraction, XRD (X´Pert PRO, PANalytical B.V., The

Netherlands). The composition of the different mineral samples is presented in Table 3.

As can be seen in Table 3, some minerals consist of a pure single phase like calcite, while

others have different minerals phases.

The biotite used in sorption and transport experiments (Paper V, VI) was sampled in

Helle, Norway and kindly provided by The Royal Museum of Natural History,

Stockholm, Sweden; (catalog number LK-5210).

19

The biotite used in the experiments was characterized elsewhere.54

Powder X-ray

diffraction analysis revealed that the specimen consisted of biotite without detectable

impurities.

The biotite specimen was cleaved manually with pincers in order to obtain fresh new

biotite particles. The particle size ranged from 1 to 3 mm in diameter. For transport

experiments (Paper VI), different amounts of biotite (10-20g) were used to pack the

columns mixed with finest quartz sand.

All the minerals were repeatedly washed with milli-Q water to remove colloidal particles

prior to the experiments.

Table 3. Structural formula, pzc and phase occurrence (%) of the minerals used in this study.

Mineral % Formula pzc

Quartz Quartz 100 SiO2 2 53

Granite Quartz 39 SiO2 2 53

Albite 31 NaAlSi3O8 2 53

Biotite 29 K(Mg,Fe)3 AlSi3O10 (OH,F)2 6.5 55

Epidote Epidote 30.7 Ca2(Fe3+

Al)3 (SiO4)3 (OH)

K-Feldspar 38 KAlSi3O8 2-2.4 53

Chlorite 21 (Mg,Fe,Al)12 (Si,Al)8 O20

(OH)16 5

56

Quartz 10.2 SiO2 2 53

Biotite 100 K(Mg,Fe)3 AlSi3O10 (OH,F)2 6.5 55

Calcite 100 CaCO3 ---

The morphology and surface roughness of the different minerals was examined by using

Scanning Electron Microscopy (SEM), using a JEOL JSM9460LV SEM/EDS

microscope, Japan. Pictures of the surface roughness of fine and coarse quartz are shown

in Figure 6.

20

Figure 6: Scanning electron microscopy image of the fine quartz sand (0.16-0.63 mm) and the

coarse quartz sand 1.4-5 mm (non-coated sample, 10 kV accelerating voltage). The scale bars is 5

µm.

The specific surface areas of the different minerals were determined by the Brunauer-

Emmett-Teller (BET) isotherm using N2 as adsorbing gas (Flowsorb II 2300,

Micromeritics, USA). The resulting specific surface areas are listed in Table 4.

Table 4. Size fraction and specific surface area of minerals.

Mineral Size fraction (mm) Specific Surface Area (m2 kg

-1)

Quartz fine 0.16-0.63 190±50

Quartz medium 0.48-1.8 340±50

Quartz coarse 1.4-5 560±50

Granite 0.5-2 410±50

Epidote 0.5-2 620±50

Calcite 0.5-2 320±50

Biotite 1-3 130±50

21

3. Methods

3.1. Single Particle Counting

Single Particle Counting (SPC) is used to count and size the particles present in water by

analyzing the particle-scattering of the laser light for a range of angles. Pulse height

analysis is used for sizing the scattered light into a number of size channels. The

instrument consists of a pump, a syringe, a laser-detector unit and a computer for data

processing.57

The HSLIS-M50 unit of the SPC measures concentration of colloidal particles in four

size channels: 50-100, 100-150, 150-200 and >200 nm. HVLIS-C200 unit measures

colloid concentrations in the larger size region, consisting of eight channels: 200-300,

300-500, 500-700, 700-1000, 1000-1500, 1500-2000, 2000-5000 and >5000 nm.58

The pump ensures a constant milli-Q water flow. The syringe is used to inject the

samples into the on-line flow and the light-scatter phenomenon is analyzed.

Given the high sensitivity of the instrument and to ensure that only one scattering event

takes place when one particle passes the laser beam, sample dilution is often needed.

3.2. Photon Correlation Spectroscopy

Photon correlation spectroscopy (PCS) measures the size of particles in solution in the

size range 10-2000 nm by determining the diffusion coefficient. The advantages of this

technique are that it is fast, non-invasive and requires only a small sample volume.

Photon correlation spectroscopy measures the dynamic light scattering (or quasi-elastic

light scattering). When a particle is illuminated by the laser beam, the phase of the

scattered light is dependent on its position. The total intensity of the scattered light is the

result of all the individual scattered waves. Fluctuations arise in the scattering intensity at

a given scattering angle because the phase and polarization of the light scattered by each

particle alter over time as the position of the particle changes due to the Brownian

motion. The fluctuations over time of the scattered light have a lifetime τ that is recorded

and analyzed using an autocorrelation function. The average of all the pulses in small

time intervals gives the intensity of the autocorrelation function C(τ):

BKDAC T )exp()( 2 (27)

where A and B are constants, DT is the translational diffusion coefficient, and K is a

constant calculated as:

)2/sin()/4( 0 nK (28)

22

where n is the refractive index of the liquid, λo is the wavelength of the laser beam and θ

is the scattering angle. Since the diffusion rate of particles is determined by their size, the

rate of fluctuation of the scattered light can be used to determine their size. The Stokes-

Einstein equation gives the relationship between the hydrodynamic diameter d and the

diffusion coefficient DT:

dTkD bT 3/ (29)

where kB is the Boltzmann constant and T is the temperature.

The signal of the PCS instrument, given as counts per second, is mainly determined by

the number of particles scattered. However, the refractive index and the geometry of the

particles can also affect the intensity of the signal.59

In this study, the PCS instrument used was a 90Plus Particle Sizer supplied by

Brookhaven Instruments Corporation. The instrument consists of a He-Ne laser source, a

set of optical elements to collimate, focus and polarize the beam, a sample cell placed in a

temperature-controlled module, a second set of optical elements to collect the scattered

light, an amplifier and a detection system that counts the number of photons occurring in

a defined time interval, a correlator that stores the counts and fits the time average

calculations to a correlation function and a computer for parameter input and data

output.60

In order to determine colloid aggregation rates, it is necessary to quantify the number of

particles in suspension. One possibility is to count each particle in suspension, for

example by using SPC, while another alternative is to determine one property of the

colloidal system that is proportional to the concentration of paticles in suspension, which

is the case when using the PCS technique.

The signal given by PCS can be used as a relative measure of the concentration of

particles under certain conditions. If the size distribution and the mean size of the

particles do not significantly change over time, the change in the count rate over time is

only dependent on the change in the concentration of particles. In a colloidal suspension,

the concentration of particles decreases with time due to aggregation and subsequent

sedimentation of large aggregates. At sufficiently low electrolyte concentrations and/or

low particle concentration, the aggregation rate is slower than the sedimentation rate.

When aggregation is the rate-determining step, the larger particles move downwards

while the smaller ones remain in the upper part of the suspension. The mean size and the

size distribution of the particles in the upper part of the suspension do not change

significantly over time. Therefore, the PCS signal, given as the count rate of the

measurement, can be used as a relative measure of the particle concentration.

23

3.3. Zeta potential analysis

The zeta potential is the value of the surface potential ψ at the Stern layer plane where the

diffuse electrical double layer of the particle starts. It can be determined from the

movement of the charged particles in the presence of an electrical field. Depending on the

sign of the charge, the particles move either to the positive or to the negative pole. The

velocity of the movement is proportional to the charge of the particle. The electrophoretic

mobility, ue, of the particles can be expressed as:

)(3

2af

E

vue

(30)

where v is the particle velocity, E is the electric field, ζ is the zeta potential in mV, ε is the

dielectric constant of the medium, η is the viscosity of the medium, f(κa) is the Henry

function, being κ the inverse thickness of the electrical double layer and a the radius of

the particles.61

The function f(κa) varies from 1 to 1.5 depending on ka. For large particles with a thin

double layer, where f(κa) is >> 1 and f(κa)=1.5, the electric field does not affect the

mobility of the particles and equation (31) is known as the Smoluchowski equation. In

the case of small particles in diluted aqueous solution, κa <1 and f(κa)=1, the ions in the

double layer surrounding the particle also move due to the electric field but in the

opposite direction to the particle, which causes a reduction in particle velocity.

Expression (31) is then known as the Hückel equation.62

A ZetaPALS Zeta Potential Analyzer supplied by Brookhaven Instruments Corporation

or a Malvern Zetasizer 3 from Malvern instruments were used to determine the zeta

potential of the particles. The acronym PALS stands for Phase Analysis Light Scattering.

Two electrodes provide an electrical field. The light of laser beam is scattered by the

particles. Since the particles are in movement, the Doppler effect of the scattered light is

used to calculate the velocity of the particles.

3.4. Fluorescence spectrophotometry

Fluorescent substances are able to absorb a photon of a given wavelength which induces

an electronic energy transition to a higher state without electron spin change. When the

excited electrons return to the ground state, the emitted radiation has usually a longer

wavelength than the radiation absorbed for excitation. The shift during emission to longer

wavelengths is known as the Stokes shift.

Qualitative and quantitative information of fluorescence substances can be obtained by

analyzing the emission spectra after excitation, considering the linear relationship

between the power of fluorescent radiation, F, and the concentration of the fluorescence

substance, c, as given by the following expression:

24

bcPKF 303.2´ 0 (31)

where P0 is the power of the incident beam, is the molar absorptivity of the

fluorescence substance, K´ is a constant depending on the quantum efficiency of the

process, and b is the length of the medium.63

The concentration of fluorescent polystyrene latex colloids was determined by means of a

Cary Eclipse Fluorescence Spectrophotometer, Varian, CA, USA. The instrument is

provided with a full spectrum Xe pulse lamp, a monochromator for excitation wavelength

selection, a second monocromator, a photoelectric detector and a computer for

operation.64

3.5. Liquid scintillation counting

In a beta decay process, the nucleus emits an electron (β- or e0

1 ), a positron (β+ or e0

1 ) or

undergoes electron capture. These three processes can be represented as:

0

11XX A

Z

A

Z (32)

0

1

0

11

0

11 eXXX A

Z

A

Z

A

Z (33)

XX A

Z

A

Z 1 (34)

where X represents the atom, A is the mass number or number of nucleons (protons and

neutrons), Z is the atomic number, v symbolizes the neutrino and represents the

antineutrino.

For low energy beta emitters (C-14, S-35, H-3) liquid scintillation is the most

appropriated detection technique. The sample containing the beta emitter is dissolved in

the scintillating solution (Ready Safe, Beckman) and placed between two photomultiplier

tubes of a beta counter Tri-Carb 1500, Packard. The energy of the beta particle is

absorbed by the scintillation liquid and light is emitted. The signal is received by the

photomultipliers and analyzed by a coincidence counter that only records when two

pulses from the photomultipliers arrive simultaneously.65

25

4. Results and Discussion

4.1. Colloid generation

It is known that the amount of colloidal particles generated increases with increasing flow

and that colloidal particles are generated at flow rates as low as 0.49 mL d-1

and under

quasi-static conditions.66,67

When a pellet of compacted bentonite is in contact with water,

the attractive and repulsive forces in the swollen montmorillonite front are in equilibrium.

The higher the swelling pressure, the more colloidal particles are generated due to higher

repulsion between aluminosilicate layers as was shown by Missana et al. 200366

If

colloidal particles are removed by flow shear forces or by self diffusion, new particles are

generated to re-establish the equilibrium.

In this thesis colloid generation in the absence of flow was studied. Under static

conditions, colloidal particles are not eroded from the source by shear force. The only

mechanism for colloid formation is the swelling of the clay, followed by particle

detachment and subsequent self-diffusion of colloidal particles from the colloid source

toward the solution.

The colloid concentration in equilibrium is the result of a cyclic process consisting of

particle detachment from the source, diffusion, aggregation, sedimentation and again

detachment. Thus, one could describe the population of colloidal particles in suspension

as dynamic since new particles are released and settle continiously. A sketch

summarizing all these processes is represented in Figure 7.

Figure 7: Colloidal life cycle.

26

Figure 8 shows photographs of generation-sedimentation experimental set-up. In the

generation experiments, the dry compacted pellet is in contact with water, and particle

detachment and diffusion take place in preference to aggregation and sedimentation. On

the other hand, in a newly-prepared colloid dispersion, aggregation and sedimentation

take place faster and the concentration of particles in suspension decreases over time.

Figure 8: Generation (a, b) and sedimentation (c, d), before (a, c) and after (b, d) the

experiments.

d c

b a

27

For both types of experiments, the final concentrations of particles in suspension are

found to be very similar, as shown in Figure 8b and 8d. This can be understood by taking

into consideration that all these processes are controlled by electrostatic forces. Thus,

when the system reaches an apparent local steady state concentration, the rate of

generation is identical to the rate of aggregation. Since the concentration of particles is

not homogeneous along the test tubes, this has been referred to as “pseudo-

equilibrium”.67

In order to quantify mass transport from the bentonite barrier under different conditions,

it is necessary to know the equilibrium concentrations of colloids outside the barrier. The

results from simple sedimentation experiments are compared to those from generation

experiments.

50-100 nm

100-150 nm

150-200 nm

200-300 nm

300-500 nm

500-700 nm

700-1000 nm

1000-1500 nm

1500-2000 nm

2000-5000 nm

Figure 9: Normalised Na-montmorillonite colloid concentrations for generation (a) and

sedimentation (b) experiments as a function of time at NaCl concentration 0.001 M. Samples

taken in the middle of the batch.

With increasing ionic strength, the equilibrium particle concentration decreases and

pseudo-equilibrium is reached more rapidly. There are two reasons for this behavior:

first, the swelling pressure decreases with increasing ionic strength,40

which reduces

colloid generation and second, as the ionic strength increases, the double layer

compresses according to Eq. (9) and aggregation becomes faster due to less repulsion

between particles. The influence of ionic strength is discussed in more detail in section

4.2.

NaCl 0.001 M

1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

1.E+09

0 1 10 100 1000 10000Time (h)

Co

llo

id c

on

cen

tra

tio

n (

ml

-1 n

m-1)

a

Generation

1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

1.E+09

0110100100010000Time (h)

Co

llo

id c

on

ce

ntr

atio

n (

ml

-1 n

m-1)

NaCl 0.001 M

b

Sedimentation

28

Figure 10: Pseudo-equilibrium profiles in generation and sedimentation experiments at time zero

and after 7000, 6000 and 2300 hours for Na-montmorillonite in 0.001, 0.01 and 0.1 M NaCl

respectively and after 1500 hours for Ca-montmorillonite in 0.001 M NaCl.

The concentration of colloids in mg L-1

in the size range 50-5000 nm can be estimated

from the pseudo-equilibrium profiles shown in Figure 10.

Assuming spherical geometry for colloidal montmorillonite particles with a density of 1.5

g cm-3

, the mass in grams of colloids at pseudo-equilibrium can be approximated by

multiplying the density of the particles by the number of colloidal particles and the

average volume in each size fraction as expressed in equation (35):

11 i

particleiiparticlei

i

iTotal VNmNm (35)

where mTotal is the concentration of montmorillonite in the suspension expressed in g L-1

,

Ni is the number of particles per litre at pseudo-equilibrium for each size fraction as

determined by SPC, mi particle is the average mass of the colloidal particle for a given size

fraction, ρ is the particle density and Vi particle is the average volume of the particles in a

given size fraction. The radius of the particles was determined as the average of each size

fraction. Due to these approximations, the mass of colloids is a rough estimate rather than

an exact value.

In the case of Na-montmorillonite in contact with 0.001 M NaCl solution, the

concentration of particles in suspension was determined as an average of the colloid

concentrations in generation and sedimentation experiments, since this system has not

reached pseudo-equilibrium.

29

Table 5 summarises the pseudo-equilibrium concentrations of colloidal Na- and Ca-

montmorillonite for each ionic strength studied.

Table 5. Pseudo-equilibrium colloid concentrations (mg L

-1) for Na- and Ca-montmorillonite in

the size range 50-5000 nm in contact with different concentrations of NaCl solution.

NaCl (M) 0.001 0.01 0.1

Na-montmorillonite 5.2±0.5 0.5±0.1 0.2±0.1

Ca-montmorillonite 0.4±0.2 --- ---

From Table 5 we can see that colloids are present in suspension despite the relatively

high salinity in the medium. Regarding the first question in the sketch presented in Figure

2, it becomes clear that colloid generation is a relevant process that should not be

disregarded. Colloid generation may gain even greater importance in a flowing system,

where physical erosion of colloidal particles will increase the colloid population.

Even if generation experiments with Ca-montmorillonite were performed in NaCl

solutions and cation exchange reactions took place, the colloid pseudo-equilibrium

concentrations differ by one order of magnitude between Na- and Ca-montmorillonite.

This indicates that cation exchange is only partial.

4.2. Stability of colloidal suspensions

A colloidal suspension is said to be stable if the particles remain in suspension during

over a long period of time, showing low tendency to aggregate. The concept of stability is

arbitrary since it depends on time as such and the time frame of other processes in the

system.

4.2.1. Method applicability and PCS calibration

Figure 11 shows the change in the mean size of different cationic forms of

montmorillonite particles dispersed in NaCl and CaCl2 electrolyte as a function of time.

As can be seen in Figure 11, variations in the relative mean size over time are not

significant. The reason for this is that large aggregates settle and are not detected by PCS

in this experimental set-up.

30

0

50

100

150

200

250

300

350

400

450

0 10000 20000 30000 40000 50000 60000

Time (min)

Siz

e (

nm

) .

NaMont+NaCl

CaMont+CaCl2

Figure 11: Mean size of Na- and Ca-montmorillonite particles as a function of time.

The PCS signal is sensitive to temperature. In Figure 12, the signal given by the

instrument against the particle concentration is represented for three different

temperatures.

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

Concentration mg ml-1

Co

un

t R

ate

(M

cp

s

-1 )

.

Figure 12: Relationship between the PCS count rate and the concentration of bentonite particles

at pH 9 and 0C (), 20C (), and 80C ().

The count rates were linearly proportional to the concentration of particles for the three

temperatures (Figure 12). Hence the calibration curves are used to quantitatively

determine colloid concentrations. However, the signal obtained at 80ºC was lower than at

0 and 20ºC. Therefore, the readings at the highest temperature were normalized by

multiplying by the factor of proportionality between 20 and 80ºC.

31

4.2.2. Effect of the ionic strength on aggregation kinetics

One of the groundwater parameters that most strongly determine the stability of a

colloidal suspension is ionic strength. This is easy to understand bearing in mind the

dependency of the repulsive force on ionic strength according to equations (9)-(14).

To evaluate the impact of ionic strength on colloid aggregation, the kinetics of

aggregation was studied by using the data treatment described in section 4.2.1. This

method allows the determination of the rate constants for aggregation at different ionic

strengths.

Plotting the inverse of the count rate versus time gave straight lines with correlation

factors R2

≥ 0.9, which indicates that the aggregation of the colloidal particles studied

(bentonite, montmorillonite and amidine latex) follows second order kinetics, as

expected.42

The aggregation kinetics of Na- and Ca-montmorillonite dispersed in solutions with

different concentrations of NaCl or CaCl2 were studied. For example Figure 13 shows the

aggregation kinetics of montmorillonite particles at different NaCl concentrations.

Na-Montmorillonite in NaCl

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0 10000 20000 30000 40000 50000 60000

Time (min)

1/C

R (

Mcp

s

-1

) .

3*10-3 M

4*10-3 M

5*10-3 M

6*10-3 M

7*10-3 M

8*10-3 M

9*10-3 M

10-2 M

Figure 13: Inverse of the count rate versus time for Na-montmorillonite at different

concentrations of NaCl.

In Figure 13 it can be observed that increasing electrolyte concentration accelerated

aggregation kinetics. Similar results were obtained using CaCl2 as electrolyte.

The true second order rate constants were estimated from the second order slopes,

expressed in Mcps-1

min-1

units, by transforming the signal of the PCS instrument, given

as count rate, to particle concentration. The number of particles in a suspension can be

expressed as:

particleTotal mNm (36)

32

where mTotal is the mass of montmorillonite in the suspension expressed in g L-1

, mParticle

is the mass of each colloidal particle and N is the number of particles per litre. For the dry

mass of montmorillonite present in the experiments mTotal the count rate given by PCS

can be related to the number of particles N and the mParticle which was deduced from the

density once the volume of the particles was known. The volume of the particles was

estimated from the particle size data obtained from PCS assuming spherical geometry and

monodisperse suspensions (Paper II). The average diameter of Na-montmorillonite

particles was found to be 220 nm, while that of Ca-montmorillonite particles was 380 nm

(Figure 11). A density of 1500-1600 kg m-3

for hydrated particles of Na-montmorillonite

and 1700-1800 kg m-3

for Ca-montmorillonite reported by Pusch (2001)68

were used in

the calculations. The experimentally determined second order slopes and the

corresponding estimated second order rate constants are summarized in Table 6 for NaCl

electrolyte and in Table 7 for CaCl2 electrolyte.

Table 6. Second order slopes and second order rate constants for the aggregation kinetics of Na-

and Ca-montmorillonite particles at different concentrations of NaCl.

NaCl (M)

Na-montmorillonite Ca-montmorillonite

Slope (Mcps-1

min-1

) × 105

k (l mol-1

s-1

)

× 10-5

Slope (Mcps-1

min-1

) × 105

k (l mol-1

s-1

) ×

10-5

3×10-3

3.2 ± 0.4 0.09 ± 0.01 13 ± 1 4.6 ± 0.4

4×10-3

4.5 ± 0.3 0.118 ± 0.008 16 ± 2 5.7 ± 0.9

5×10-3

4.9 ± 0.2 0.134 ± 0.006 19 ± 2 6.0 ± 0.6

6×10-3

5.4 ± 0.2 0.147 ± 0.006 29 ± 1 8.2 ± 0.3

7×10-3

6.4 ± 0.4 0.172 ± 0.05 40 ± 5 11 ± 1

8×10-3

7.1 ± 0.2 0.194 ± 0.007 50 ± 10 14 ± 3

9×10-3

15.8 ± 0.7 0.43± 0.02 80 ± 20 19 ± 5

1×10-2

23.3 ± 0.9 0.63 ± 0.02 90 ± 10 23 ± 3

Table 7. Second order slopes and second order rate constants for the aggregation kinetics of Na-

or Ca-montmorillonite particles at different concentrations of CaCl2.

CaCl2 (M)

Na-montmorillonite Ca-Montmorillonite

Slope (Mcps-1

min-1

) × 105

k (l mol-1

s-1

) ×

10-5

Slope (Mcps-1

min-1

) × 105

k (l mol-1

s-1

) ×

10-5

3×10-4

4.5± 0.2 0.163 ± 0.009 22.1 ± 0.8 11.2 ± 0.4

4×10-4

7.3± 0.4 0.31 ± 0.02 47 ± 2 24 ± 1

5×10-4

26 ± 3 0.9 ± 0.1 54 ± 8 27 ± 4

6×10-4

37 ± 3 1.3 ± 0.1 103 ± 15 52 ± 8

7×10-4

155 ± 48 5 ± 2 166 ± 29 80 ± 14

8×10-4

218 ± 39 8 ± 1 197 ± 39 100 ± 20

9×10-4

400 ± 132 14 ± 5 276 ± 23 140 ± 9

As can be seen in Tables 6 and 7, the rate constants for aggregation increase with

increasing electrolyte concentration for both cationic forms of montmorillonite and

electrolytes, in agreement with DLVO theory.

33

When the data listed in Table 6 and Table 7 was plotted as the logarithm of the second

order rate constants versus the square root of the ionic strength of the suspensions, a

linear relationship was obtained (Figure 14).

Figure 14: Effect of electrolyte concentration on the aggregation kinetics of Na- and Ca-

montmorillonite for a) NaCl and b) CaCl2. CE indicates the electrolyte concentration necessary to

replace 98% of the exchangeable cations in montmorillonite.

In Figure 14, the ionic strength denoted by CE indicates the electrolyte concentration

required to exchange 98% of the calcium in Ca-montmorillonite dispersed in NaCl for

sodium (Figure 14a), and the concentration of CaCl2 necessary to replace 98% of the

sodium in Na-montmorillonite for calcium (Figure 14b). The CE quantities were

calculated from the cation exchange capacity of montmorillonite18

using the equilibrium

constants reported by Tang et al. 199369

for 98% replacement of the exchangeable

cations. Since cation exchange is an equilibrium reaction, a larger degree of replacement

occurs in the interlayer space as the concentration of electrolyte in the medium increases.

In Figure 14a, the calculated CE is lower than the concentrations used in the experiments.

Therefore, replacement of calcium by sodium in Ca-montmorillonite is expected to be

y = 91.821x + 1.379 R 2 = 0.984

y = 49.292x + 4.601 R 2 = 0.983

0

1 2

3

4

5

6 7

8

0 0.01 0.02 0.03 0.04 0.05 0.06

I 1/2

Lo

g k

Na-Montmorillonite

Ca-montmorillonite

CE= 0.26 b

y = 16.765x + 2.9556

R 2 = 0.8351

y = 16.043x + 4.7206 R2 = 0.9653

0

1

2

3

4

5

6

7

0 0.02 0.04 0.06 0.08 0.1 0.12

I 1/2

Lo

g k

Na-Montmorillonite

Ca-Montmorillonite

CE a

34

complete for the range of NaCl concentrations used in the experiments. Since both

materials are Na-montmorillonite, they show the same dependency on the logarithm of

the rate constant with the square root of the ionic strength. However, in the experiments

performed with CaCl2 (Figure 14b), Na- and Ca-montmorillonite showed different rates

of increase in log (k) with increasing ionic strength. The calculated CE value is much

higher than the CaCl2 concentrations used in the experiments. Thus, cation exchange

reactions can be expected to take place in the range of ionic strengths investigated. A

larger replacement of sodium by calcium in Na-montmorillonite occurred as the

concentration of CaCl2 increased.

The zeta potential of montmorillonite at buffer pH and room temperature was

independent of electrolyte concentration, but strongly dependent on the cationic form of

montmorillonite and the cation of the electrolyte. The average zeta potential values of

different electrolyte concentrations are given in Table 8.

Table 8. Average values of zeta potential of Na- and Ca-montmorillonite in de-ionized water and

NaCl or CaCl2 electrolyte.

Deionised water

(mV) NaCl (mV) CaCl2 (mV)

Na-montmorillonite -55 ± 2 -54 ± 9 -25 ± 5

Ca-montmorillonite -18 ± 2 -47 ± 8 -19 ± 4

The valence of the cation adsorbed in the Stern layer strongly determines the zeta

potential of the particles. Consequently, the charge of the montmorillonite colloid is

completely governed by the electrolyte in the medium.

DLVO calculations were performed in order to interpret the experimental results from

aggregation kinetics studies. The differences in zeta potential for the pure sodium and

pure calcium systems were taken into account when calculating the repulsive, attractive

and total energy using equations (9) to (14), assuming the zeta potential values to be

proportional to the surface potential of the particles.

The maximum values reached by the total energy are plotted against the square root of

the ionic strength in Figure 15.

35

y NaCl = -6.494E-18x + 4.470E-18

R2 =0.998

yCaCl2 = -8.513E-18x + 2.616E-18

R2 = 0.998

0

5E-19

1E-18

1.5E-18

2E-18

2.5E-18

3E-18

3.5E-18

4E-18

4.5E-18

0 0.05 0.1 0.15 0.2

I ½

VT

ma

x (

J)

.

NaCl

CaCl2

Figure 15: Maximum values of the total energy versus square root of the ionic strength. The solid

lines show linear extrapolations of the values calculated at the low ionic strengths inside the

square.

As can be seen in Figure 15, the height of the total energy function decreased with

increasing electrolyte concentration, which is in agreement with the kinetic results. For a

limited ionic strength range, such as the ionic strength range investigated experimentally

marked by a square in Figure 15, the maximum of the function decreases linearly with the

square root of the ionic strength. However, it deviates from linearity for a wider ionic

strength range. This explains the linear correlation between the logarithm of the rate

constant and the square root of the ionic strength observed for the fairly narrow ionic

strength range investigated experimentally (Figure 14a and 14b). Note that the total

energy maximum is more sensitive to increasing CaCl2 than increasing NaCl ionic

strength, which is in agreement with experimental results shown in Figure 14.

The linear relationships shown in Figure 14a and 14b allowed CCC to be estimated by

extrapolation to diffusion-controlled conditions. The diffusion-controlled rate constant

for particle aggregation calculated from equations (6)-(8), 6.53 x 109 l mol

-1 s

-1, was

inserted into the linear regression for the log (k) with I obtained experimentally for the

pure sodium and calcium systems. The ionic strengths obtained in this way correspond to

apparent CCC values. However, since log (k) is expected to be linearly related to VT max

and VT max is not strictly linearly related to I , the apparent CCC values must be

corrected. The correction is performed by identifying the I value at which the VT max

value corresponds to the diffusion-controlled limit using DLVO theory. The correction

procedure is illustrated in Figure 16 for the case of CaCl2.

36

y = -8.513E-18x + 2.616E-18

R2 = 0.998

0

5E-19

1E-18

1.5E-18

2E-18

2.5E-18

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

I ½

VT

ma

x (

J)

. .

CaCl2

CCC

corrected

CCC

apparent

Figure 16: Correction of the CCC value for Ca-montmorillonite in CaCl2 electrolyte.

The corrected CCC for Na-montmorillonite in NaCl was found to be 0.70 M and the

corrected CCC for Ca-montmorillonite in CaCl2 solutions was 0.0062 M. These CCC

values are of the same order of magnitude or higher than the CCC values published in the

literature, as can be seen in Table 9.

Table 9. Published CCC values for montmorillonite compared with those obtained in the present

study (Paper II).

Author (year) Mineral Clay/water

(%) Method CCC (M)

Frey (1978)71

Wyoming 0.025 Test tube series

(24h) 0.01-0.15 NaCl

Tombácz (1989)71

Na-Mont 0.2 Visual inspection

(24h)

0.25 NaCl

0.002 CaCl2

Tombácz(1989) 71

Ca-Mont 0.2 Visual inspection

(24h)

0.25 NaCl

0.002 CaCl2

Chheda (1992) 72

Na-Mont 0.015 Turbidity and

optical density

≈ 0.018 NaCl

≈0.00045 CaCl2

Lagaly (2002) 73

Na-Mont 0.025 Visual inspection 0.005 NaCl

0.0004 CaCl2

Lagaly (2002) 73

Na-Mont 0.5 Rheological

measurements

0.015 NaCl

0.002 CaCl2

Lagaly (2002)73

Na-Mont 1 Rheological

measurements

0.020 NaCl

0.003 CaCl2

SKB (2004a) 74

Mont Not

specified Not specified

0.1 NaCl

0.001 CaCl2

Paper II Na-Mont 0.08 Aggregation

kinetics 0.70±0.05 NaCl

Paper II Ca-Mont 0.08 Aggregation

kinetics

0.0062±0.0005

CaCl2

37

The CCC values determined by aggregation kinetics can be expected to be higher, since

other experimental methods (listed in Table 9) identify the CCC value with changes in

the colloidal system that become significant at aggregation rate constants far below the

point where the system begins to be diffusion-controlled. It is interesting to note that

under diffusion-controlled conditions, the total energy is still positive, when according to

the CCC definition given by DLVO theory the total energy should be zero. Therefore,

according to the DLVO theory, the CCC value should correspond to an even higher

electrolyte concentration.

The difference between the CCC values for NaCl and CaCl2 of two orders of magnitude

is mostly due to the differences in zeta potential, size and density between the Na- and

Ca-montmorillonite particles. However it would be of interest to use this method to

determine CCC values for colloidal systems not affected by cation exchange processes.

4.2.3. Influence of temperature on aggregation kinetics

The effect of temperature on the stability of bentonite colloids at pH≈9 and amidine latex

particles dispersed in NaCl or NaClO4 solutions (pH=5.4) was investigated. The second

order kinetics for bentonite particles at 20 and 80ºC are shown in Figure 17.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

0 1000 2000 3000 4000 5000 6000 7000 8000

Time (min)

1/C

R (

Mcp

s

-1

) .

Figure 17: Inverse of the count rate as a function of time at 20C () and 80C () for bentonite

suspensions in 10-3

M NaCl or NaClO4 solutions at pH 9.

The second order slopes were the same for 10-3

M NaCl and NaClO4. At 20ºC the second

order slopes for both electrolytes were (10.0±0.1) × 10 -5

Mcps-1

min-1

and at 80ºC the

second order slopes were (2.7±0.5) × 10 -5

Mcps-1

min-1

. The faster aggregation kinetics

at 20ºC indicates lower stability at 20ºC compared with that at 80ºC.

However, the effect of temperature on the stability of amidine latex particles was the

opposite of that for bentonite. Increasing temperature had a destabilizing effect on

38

amidine latex colloids dispersed in 10-3

M NaCl or NaClO4 solutions, as can be seen in

Figure 18.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 20 40 60 80 100 120 140

Time (min)

1/C

R (

Mcp

s

-1)

.

Figure 18: Sedimentation kinetics for amidine latex particles in 10

-3 M NaCl or NaClO4

electrolyte at pH 5.7, 20C () and pH 5.4, 80C ().

Suspensions at 20C were more stable as shown by the second order slope (1.6±0.2∙10 -5

Mcps-1

min-1

), which was lower than the slope at 80C (3.7 ±0.5∙10 -5

Mcps-1

min-1

).

The zeta potential of bentonite particles at pH≈9 was independent of temperature and

stable over time as can be seen in Figure 19.

-60

-55

-50

-45

-40

-35

-30

0 1000 2000 3000 4000 5000 6000 7000 8000

Time (min)

Zeta

po

ten

tial

(mV

)

.

Figure 19: Zeta potential of bentonite colloids in 10

-3 M NaCl or NaClO4 electrolyte at pH 9 as a

function of time at 0C (), 20C (), and 80C ().

No differences in the zeta potential were observed between NaCl and NaClO4. Table 8

shows that that the charge of the positive counter-ions strongly determined the zeta

39

potential of the particles. Since the cation in both electrolytes was monovalent Na+, the

zeta potential of bentonite particles should be the same. If the bulk composition is not

changed, the Stern layer does not change and neither does the zeta potential. Therefore,

the zeta potential should be constant with time.

Increasing temperature had a dramatic effect on the zeta potential of amidine latex

particles, as can be seen in Figure 20.

0

10

20

30

40

50

60

70

80

90

0 1000 2000 3000 4000 5000 6000 7000 8000

Time (min)

Zeta

po

ten

tial

(mV

)

.

Figure 20: Zeta potential of amidine latex colloids in 10

-3 M NaCl or NaClO4 electrolyte as a

function of time at pH 5.7, 20C (), and pH 5.4, 80C ().

Amidine latex particles are positively charged, and the Stern layer consists of the

negative counter-ions of the electrolyte in solution. No significant differences in zeta

potential were observed for the anions Cl- and ClO4

-, since both anions are monovalent.

The small influence that the different size of the anions could have on the zeta potentials

measurements, is probably of the same magnitude as the error.

Increasing temperature induces the deprotonation of amidine head groups NH2+(NH2).

75

Therefore, the zeta potential was less positive at the higher temperature (Figure 20). This

is in agreement with the fact that the pH of the suspensions at 80°C was slightly more

acidic than the pH of the suspensions at 20°C. Note that the zeta potential of amidine

particles at 80°C slightly decreased over time, indicating progressive deprotonation of the

head groups. The lower zeta potential at 80°C explains the faster aggregation kinetics

observed experimentally (Figure 18).

When the DLVO theory, equations (9)-(14), was used to analyze the effect of

temperature on the stability of bentonite particles, it showed that the repulsive energy,

and in consequence the total energy, increased with increasing temperature, as can be

seen in Figure 21.

40

-8E-18

-6E-18

-4E-18

-2E-18

0

2E-18

4E-18

6E-18

8E-18

0 1 2 3 4 5 6 7 8 9 10

Distance (nm)

VR,

VA,

VT (

J)

VA

VT

VR

Figure 21: Calculated energy curves of interaction (in J) as a function of interparticle distance

and temperature at pH 9 for bentonite particles in 10-3

M of 1:1 electrolyte. The arrows indicate

the evolution of the potential functions with increasing temperature.

These results are in agreement with the aggregation kinetics, where increasing

temperature was shown to have a stabilizing effect on bentonite suspensions. As the

maximum of the total energy increases with increasing temperature, a smaller fraction of

particles overcomes the energy barrier and the aggregation process becomes slower.

However, when the effect of temperature on the total energy was investigated for amidine

latex, the calculations showed that the total energy decreases with increasing temperature

(Figure 22).

41

-8E-18

-6E-18

-4E-18

-2E-18

0

2E-18

4E-18

6E-18

8E-18

0 1 2 3 4 5 6 7 8 9 10

Distance (nm)

VR,

VA,

VT (

J)

VR 20°C

VT 20°C

VR 80°C

VT 80°C

VA

Figure 22: Total energy curves of interaction (in J) as a function of distance and temperature for

amidine latex particles in 10-3

M of 1:1 electrolyte.

The DLVO calculations explained the faster aggregation kinetics of amidine latex

suspensions at 80°C compared with 20°C. The reduction in the surface charge by

deprotonation of the amidine groups at 80°C drastically reduced the repulsion of the

particles. The energy barrier for aggregation was lower at the highest temperature and the

particles aggregated faster.

4.2.4. Ionic strength, pH and temperature effects on the stability of

montmorillonite colloids

Since ionic strength, temperature and surface potential were found to affect the total

energy and considering that the surface charge of montmorillonite edge groups is pH

dependent, the study of the effects of temperature and ionic strength on the stability of

montmorillonite colloids was extended to other pH values.

The aggregation kinetics for montmorillonite suspensions were studied in the pH range 3

to 11. The temperatures investigated were 4, 22 and 70°C and the ionic strength was

varied from 0.002 to 0.0068 M NaCl. The second order slopes obtained are presented in

Table 10.

42

Table 10. Second order slopes for montmorillonite aggregation kinetics expressed in Mcps-1

min-1

From Table 10 it can be seen that:

- Increasing ionic strength accelerated the aggregation kinetics of montmorillonite

particles for all the temperatures and pH values studied. This was due to a decrease in the

effective repulsion between the charged colloidal particles with increasing ionic strength.

- The aggregation rate constant decreased with increasing pH. This effect became more

pronounced at higher ionic strengths and higher temperatures but was not observed at

4ºC.

- The effect of temperature was pH- and ionic strength-dependent. As the temperature

increased, the suspensions at pH≤4 were dramatically destabilized, regardless of ionic

strength. At pH≥10 the aggregation rate constant decreased with increasing temperature.

In the intermediate pH interval, the effect of increasing temperature depended on the

ionic strength.

These experimental observations can be explained by the DLVO theory. The DLVO

theory was used to calculate the maximum of the total energy for the different

temperatures, pH and ionic strength used in the experiments. The relationship is shown in

Figure 23.

Ionic

Strength

(M)

T

(°C)

pH 3

(Mcps-1

min-1

) x 105

pH 4

(Mcps-1

min-1

) x 105

pH 6

(Mcps-1

min-1

) x 105

pH 8.3

(Mcps-1

min-1

) x 105

pH 10

(Mcps-1

min-1

) x 105

pH 11

(Mcps-1

min-1

) x 105

0.002 4 7 ± 1 7.3 ± 0.9 7 ± 1 8 ± 1 9.1 ± 0.8 12 ± 1

0.002 22 10 ± 3 9 ± 2 11 ± 2 11 ± 2 12 ± 2 18 ± 4

0.002 70 280 ± 60 60 ± 30 3.7 ± 0.9 2.8 ± 0.6 2.7 ± 0.4 11 ± 3

0.003 4 9.9 ± 0.7 11.0 ± 0.7 14.2 ± 0.9 12.2 ± 0.9 12 ± 1 20 ± 2

0.003 22 130 ± 20 17 ± 1 14 ± 4 13 ± 3 16 ± 1 24 ± 3

0.003 70 600 ± 100 250 ± 20 13 ± 2 12 ± 2 14 ± 3 19 ± 4

0.005 4 15 ± 5 12 ± 3 19 ± 2 13 ± 2 13 ± 5 ---

0.005 22 290 ± 60 38 ± 5 21 ± 2 20 ± 2 27 ± 2 ---

0.005 70 600 600 16 ± 1 16 ± 3 19 ± 4 ---

0.0068 4 24 ± 2 22 ± 5 27 ± 5 24.6 ± 0.6 25 ± 5 31 ± 5

0.0068 22 600 43 ± 4 33 ± 2 28 ± 5 34 ± 6 40 ±3

0.0068 70 600 600 600 500 ± 200 22 ± 1 26 ± 3

43

0

-100

-200

-300

-400 0

0.002

0.004

0.006

0.008

-2

0

2

4

6

8

x 10-18

Ionic strength (M)Surface potential (mV)

VT

max

(J)

0

1

2

3

4

5

6

7

x 10-18

4°C

22°C

70°C

Figure 23: Maxima of the total interaction energy as a function of temperature, pH and ionic

strength.

Ionic strength determines the thickness of the double layer, equation (9). As the ionic

strength increases, the thickness of the double layer decreases and the repulsion between

particles declines. This can be related to the observed faster aggregation with increasing

ionic strength (Paper II).

Decreasing the pH induces protonation of the surface groups.76,77,78

The γ parameter

accounts for the surface potential in the repulsive energy expression in accordance with

equation (12). Since the repulsive energy is directly proportional to the square of γ, a

decrease in the surface potential reduces the repulsive energy and thus the particles

aggregate faster.

Increasing temperature increases the collision frequency and the kinetic energy of the

colloidal particles. In general, this leads to faster aggregation as observed at pH≤4 (Table

10). However, the temperature also affects the repulsion between the particles in two

opposing ways. In Figure 24, the difference between the calculated maxima of the total

energy functions at 22 and 70ºC, i.e. (Vmax70

-Vmax22

), is represented as a function of the

absolute value of the surface potential ψo.

44

-2,0E-19

0 100 200 300 400 500 600

Vm

ax

70-V

max

22

(J)

.

Surface Potential (mV)

0.002 M

0.0068 M

3.0E-19

-1.0E-19

0.0

2.0E-19

Figure 24: Difference between the calculated total energy maxima at 70 and 22ºC as a function

of the surface potential for two ionic strengths, 0.002 M and 0.0068 M NaCl.

In Figure 24 it can be seen that the direction of the temperature effect depended on the

surface potential and that the magnitude of the effect depended on the ionic strength. At

low surface potentials the repulsion decreased with increasing temperature while at high

surface potentials the repulsion increased with increasing temperature.

The increase in the repulsion with increasing temperature counteracted the temperature-

induced increase in collision frequency. Therefore a stabilization effect was observed

with increasing temperature in the suspensions in the intermediate and high pH region

(Table 10).

The parameter γ decreases with increasing temperature, equation (12). This effect alone

would lead to weaker repulsion at higher temperature. However, the magnitude of the

temperature effect on γ decreases dramatically with increasing surface potential. The

ionic medium contribution to the repulsion (ionic strength-dependent terms) is also

temperature-dependent. Contrary to γ, the medium contribution increases with increasing

temperature and decreases with increasing ionic strength. The temperature effect on the

medium contribution counteracts the effect on the surface charge contribution (γ). Hence,

for low surface potentials (low pH) the temperature effect can be mainly attributed to the

effect on the surface charge contribution (γ) while for higher potentials (high pH) the

temperature effect can be attributed to the ionic medium contribution to the repulsion.

Concerning the second question in Figure 2 referring tocolloid stability, colloid

aggregation is affected by a number of different parameters. In the context of

groundwater, ionic strength is the most important since it has the strongest impact and

shows greater variability in fracture formations than pH and temperature. Temperature,

however becomes relevant in the near field of deep geological repositories.

45

4.3. Adsorption of colloids on mineral surfaces

In analogy with colloid-colloid interactions, colloidal particles may also have the ability

to interact with the fracture surface minerals. One way to evaluate the extent of colloid

sorption is to determine distribution coefficients (Kd) between solution and minerals by

batch sorption experiments.

The conventional way to express Kd is based on the sorbent mass but a more adequate

and general way of expressing the Kd value is by normalizing it to the exposed surface

area as given in expression (29). This allows for more relevant comparison between

sorption capacities of different solids independent of particle size.

Based on measurements of the fraction of sorbate remaining in solution as a function of

the amount of solid substrate (sorbent), a linear form of equation (24) can be employed.

Kd is obtained from the slope when plotting f

fi

C

CC against

V

sm a.

The Kd values of fluorescent polystyrene latex colloids with diameters 50, 100, and 200

nm diameter, were determined in 30 mL colloidal suspensions in contact with quartz,

granite, and epidote, biotite and calcite. The values are summarized in Table 11. For

calcite, only Kd values at pH 8.5, were determined since the mineral dissolves at lower

pH.79

The sorption of latex colloids is strongly dependent on the buffer concentration and pH of

the medium as well as the type of mineral as can be seen in Table 11. These observations

suggest that sorption of colloids on mineral surfaces is governed by electrostatic forces.

Similar results were obtained for gold colloid sorption on granite and latex colloid

sorption on different minerals in the Grimsel granodiorite matrix.80,81,82

The data presented in Table 11 does not indicate any effect of particle size on colloid

sorption, rather that sorption appears to depend on colloid surface charge. The zeta

potential of 50 nm latex particles is significantly lower (less negative) than the

corresponding values for 100 and 200 nm particles (Table 2), which implies that the

electrostatic barrier between 50 nm latex colloids and mineral surfaces should be lower.

This explains the higher Kd values for 50 nm particles. The electrostatic barrier further

reduces at high ionic strength and low pH, resulting in higher Kd values for all particle

sizes, according to DLVO expressions (9-14).

Among all mineral studied, biotite and calcite show markedly higher Kd values compared

to quartz, granite, and epidote. This trend can be related to the trend in point of zero

charge (pzc) of the minerals (Table 3). A mineral showing a higher pzc, has more

positive surface charges in the pH range studied, which favours sorption of negatively

charged latex particles.

46

Table 11. Distribution coefficients (Kd) for latex colloidal particles.

Mineral,

pH

Coll size

(nm)

Kd 0.001M

(m)x106

Kd 0.01M

(m)x106

Mineral,

pH

Coll size

(nm)

Kd 0.001M

(m)x106

Kd 0.01M

(m)x106

Quartz, 8.5 50 1.2 ± 0.4 4.0 ± 0.6 Epidote, 8.5 50 1.8 ± 0.2 6 ± 1

100 0.8 ± 0.1 3.0 ± 0.7 100 0.38 ± 0.07 1.2 ± 0.2

200 0.5 ± 0.1 1.0 ± 0.2 200 0.5 ± 0.1 1.5 ± 0.2

Quartz, 6 50 1.2 ± 0.2 16 ± 2 Epidote, 6 50 1.7 ± 0.2 7 ± 1

100 0.4 ± 0.1 11 ± 3 100 0.33 ± 0.06 5 ± 2

200 0.55 ± 0.05 12 ± 1 200 0.5 ± 0.1 4 ± 1

Quartz, 4.5 50 1.7 ± 0.4 17 ± 3 Epidote, 4.5 50 2.4 ± 0.2 10 ± 2

100 1.7 ± 0.1 20 ± 4 100 0.9 ± 0.2 8 ± 2

200 1.3 ± 0.3 20 ± 3 200 0.8 ± 0.1 9 ± 2

Granite, 8.5 50 0.4 ± 0.1 3 ± 1 Biotite, 8.5 50 23 ± 2 94 ± 10

100 0.40 ± 0.04 1.1 ± 0.2 100 9 ± 2 33 ± 3

200 0.3 ± 0.1 0.92 ± 0.2 200 17 ± 3 24 ± 5

Granite, 6 50 0.8 ± 0.1 13 ± 2 Biotite, 6 50 60 ± 20 270 ± 80

100 0.7 ± 0.1 11 ± 1 100 22 ± 2 280 ± 80

200 0.7 ± 0.1 6 ± 1 200 23 ± 5 340 ± 80

Granite, 4.5 50 0.7 ± 0.1 11.1 ± 0.7 Biotite, 4.5 50 61 ± 8 260 ± 10

100 0.3 ± 0.1 20 ± 2 100 60 ± 10 330 ± 50

200 0.9 ± 0.2 12.3 ± 0.2 200 29 ± 4 300 ± 50

Calcite, 8.5 50 25 ± 6 37 ± 5

100 7 ± 1 22 ± 6

200 5.1 ± 0.5 10 ± 1

The Kd values for sorption of latex colloids on granite at 0.001 M given in Table 11 and

gold colloids reported by Alonso et al., 200980

are comparable. The similarity of the Kd

values of the two types of colloids may be attributed to the fact that the zeta potential of

gold colloids and the specific surface area of granite are quite similar in both studies.

Similar Kd values were also obtained for granite and epidote (Table 11). These minerals

are also expected to have similar pzc values, judging from the properties of the individual

minerals composing the matrix. Considering the minerals with high pzc (chlorite and

biotite), the Kd values of granite and epidote could be expected to be higher than that of

quartz. As can be seen in Table 11, this was not the case. Most likely, it is due to non-

additive effects of individual constituents on the overall surface properties of granite and

epidote. Since in biotite faces and edges have different properties, it is important which

47

parts are exposed in fracture filling minerals for the sorption. On the other hand, in

sorption experiments all surfaces are exposed.

The Kd values of Na-montmorillonite colloidal suspension in contact with quartz sand at

pH 8.5, 6, or 4.5 and buffer concentration 0.001 M are presented in Table 12.

Table 12. Distribution coefficients (Kd) for Na-montmorillonite colloidal particles in 0.001 M

buffer concentration.

Mineral pH Colloid size (nm) Kd (m) x106

Quartz 8.5 20-600 1.6 ± 0.2

Quartz 6 20-600 2.2 ± 0.2

Quartz 4.5 20-600 2.4 ± 0.2

As can be seen in Table 12, similar to latex colloids, the Kd values of Na-montmorillonite

increase with increasing pH, since the positive surface charges increases, and thus

repulsion is reduced, both for quartz and Na-montmorillonite.

Considering the process of colloid sorption onto minerals (question nr.3 in Figure 2),

colloids will be partly retained in fractures by chemical interactions.

Despite the unfavourable conditions for sorption between negative colloids and surfaces

and repulsive forces dominate between colloids and surfaces, sorption is indeed observed.

4.4. Colloid transport

So far, individual processes contributing to colloid facilitated transport have been studied.

This section deals with colloid transport in which several of the individual processes are

involved.

For all the experiments performed with polystyrene latex particles, the breakthrough

curves are symmetrical and the three colloidal sizes break simultaneously with each other

and the water front. The recovery of 50 nm colloids is significantly lower than the

recoveries of the 100 and 200 nm particles. Figure 25 shows a representative

breakthrough curves and recovery curve of transport experiments performed with latex

colloids.

48

NaCl 0.001M, Porosity 0.4, 0.03 mL/min

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Pore volumes

Ct/

Ni (m

L -1 )

.

50 nm, pH=6

100 nm, pH=6

200 nm, pH=6

50 nm, pH=8.5

100 nm, pH=8.5

200 nm, pH=8.5

NaCl 0.001M, Porosity 0.4, 0.03 mL/min

0

20

40

60

80

100

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Pore volumes

Recovery

(%

)

.E

50 nm, pH=6

100 nm, pH=6

200 nm, pH=6

50 nm, pH=8.5

100 nm, pH=8.5

200 nm, pH=8.5

HTO

Figure 25: Breakthrough and recovery curves for latex particles of 50, 100 and 200 nm diameter

in quartz transport experiments at pH 6 and 8.5 with 40% porosity and flow rate 0.03 mL min-1

.

As can be seen in Figure 25, pH has no effect on colloid transport within the limited pH

range used in the experiments. This is expected considering that, for this pH range, no

significant changes on the surface charge of quartz would take place since the pzc of

quartz is around 2, and the zeta potential of the colloidal particles is not affected

significantly according to Table 2.

Colloid recovery is always lower than 100%, and significant colloid retention takes place

in the columns. Considering that the breakthrough curves are fairly symmetric, and that

the recovery curves reach a plateau after a given time, it indicates that the immobilization

of latex particles in the column is irreversible for the time frame of the experiments. The

results from all transport experiments are summarized in Table 13.

49

Table 13. Experimental conditions and results of transport experiments.

As expected, colloid recovery increases with increasing flow rate. As the flow rate

decreases, porosity starts to have an effect on colloid recovery.

It is also interesting to note that the recovery increases with increasing particle size at the

lowest flow rate. This is not the general trend at higher flow rates.

According to the Kd values presented in Table 11, latex particles more strongly sorb to

biotite than to quartz given the lower pzc for biotite compared to quartz.

Exp. No. Flow rate

(mL min-1

)

Porosity

(%)

Column

mass

(kg)

pH/

NaCl

(M)

Colloid

Type/size

(nm)

Recovery

(%)

1 0.60±0.05 Quartz/40 0.130 8.5/

0.001

Latex 50 50±3

Latex 100 92±4

Latex 200 79±3

2 0.030±0.004 Quartz/40 0.130 8.5/

0.001

Latex 50 16±2

Latex 100 56±2

Latex 200 56±2

3 0.030±0.004 Quartz/40 0.130 6/ 0.001

Latex 50 20±2

Latex 100 56±2

Latex 200 60±2

4 0.0020±0.0002 Quartz/40 0.130 8.5/

0.001

Latex 50 15±1

Latex 100 48±2

Latex 200 50±2

5 0.60±0.05 Quartz/33 0.140 8.5/

0.001

Latex 50 52±2

Latex 100 93±4

Latex 200 69±3

6 0.030±0.004 Quartz/33 0.140 8.5/

0.001

Latex 50 35±2

Latex 100 85±4

Latex 200 71±3

7 0.0020±0.0002 Quartz/33 0.140 8.5/

0.001

Latex 50 6±1

Latex 100 31±1

Latex 200 36±1

8 0.0020±0.0002

Quartz+

Biotite

/34

0.127+

0.010

8.5/

0.001

Latex 50 2±1

Latex 100 14±1

Latex 200 24±1

9 0.0020±0.0002

Quartz+

Biotite

/35

0.124+

0.020

8.5/

0.001

Latex 50 2±1

Latex 100 9±1

Latex 200 12±1

10 0.0020±0.0002 Quartz/33 0.140 8.5/

0.001

Mont. 50-100 30±4

Mont.100-150 33±5

Mont.150-200 40±6

11 0.0020±0.0002 Quartz/33 0.140 8.5/

1-0.001

Mont. 50-100 22±3

Mont.100-150 23±3

Mont.150-200 28±4

50

In order to assess the impact of sorption on colloid retention, experiments using columns

packed with quartz mixed with two different amounts of biotite were performed at the

lowest flow rate. The resulting recovery curves are shown in Figure 26.

0.002 mL/min, 0.33-0.35 porosity

0

5

10

15

20

25

30

35

40

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Pore volumes

Recovery

(%

)

.

200 nm, quartz

200 nm, 10g biotite

200 nm, 20g biotite

Figure 26: Comparison of the recovery curves for latex colloids 200 nm diameter in transport

experiments at flow 0.002 mL min-1

using 140 g clean fine quartz and 127 or 124 g quartz mixed

with 10 and 20 g free biotite respectively.

Although changes on the pore structure may have been induced by introducing biotite in

the columns, the recovery trend appears to reflect the increasing sorption capacity of the

column. This observation and the fact that the relative recovery trends in Table 11 at the

lowest flow rates in pure quartz columns reflect the Kd values of the different particle

sizes point in the direction that sorption is of significance at low enough flows. At higher

flows however, the relationship between recovery and Kd values is less obvious,

indicating that at fast flow, sorption equilibrium is not reached. It should be noted that the

impact of physical filtration would also decrease with increasing flow.

The previous observations can be better illustrated by plotting the recovery against the

total sorption capacity (filling material mass x specific surface area x Kd) of the columns

for the different experiments as shown in Figure 27.

51

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100

0.002 ml/min, quartz 33% porosity

0.002 ml/min, quartz + 10 g biotite, 34 % porosity0.002 ml/min, quartz + 20 g biotite, 34 % porosity0.002 ml/min, quartz 40 % porosity

0.03 ml/min, quartz 33 % porosity

0.03 ml/min, quartz 40 % porosity

Sorption capacity (m3)

Figure 27: Relationship between experimental recovery and column sorption capacity.

From Figure 27 it becomes evident that at low flow where the residence time is long, and

porosity of 33% where the available surface area is larger, sorption appears to be the

dominant retention process. In the experiments with higher porosities and/or higher flow

rates, the higher recoveries observed are indicative of less interaction between the latex

particles and the mineral surfaces. The lower recoveries of the 50 nm latex particles can

be explained by higher Kd values than for the larger latex particles. The 50 nm latex

particles possess a lower negative charge than the 100 and 200 nm particles giving higher

Kd values. It should be noted that size exclusion would give the same trend, however, the

effect would most probably be larger since the difference in the particle sizes are quite

large. The resemblance in the 100 and 200 nm particles behavior, would suggest that it is

more likely that the effect is reflecting sorption with similar sorption affinities for the

mineral.

The transport of colloidal Na-montmorillonite particles was studied at the lowest flow

and porosity in columns packed with fine quartz sand. The breakthrough and recovery

curves are shown in Figure 28.

52

Na-montmorillonite, porosity 0.33, flow 0.002 ml/min

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

Pore volumes

Ct/N

i (m

L

-1 )

.

50-100 nm

100-150 nm

150-200 nm

Na-montmorillonite, porosity 0.33, flow 0.002 ml/min

0

5

10

15

20

25

30

35

40

45

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

Pore volumes

Recovery

(%

)

.

50-100 nm

100-150 nm

150-200 nm

Figure 28: Breakthrough and recoveries for the Na-montmorillonite particles in three size classes

in a quartz column with porosity of 33 % and flow of 0.002 mL min-1

.

One important difference between the breakthrough curves for the Na-montmorillonite

particles compared to latex colloids is that the former are not symmetrical. The tailings in

the breakthrough curves indicate reversible retention, while the retention in the transport

of latex colloids appears to be irreversible, at least for this experimental time frame.

Another possibility is that small colloids are generated from the larger aggregates trapped

in the column. The released smaller particles are then able to be transported.

53

5. Concluding remarks

Concentration of montmorillonite colloids outside the barrier can be predicted

from the results of laboratory experiments on generation and sedimentation of

montmorillonite colloids. The concentrations obtained from the two different

types of experiments are identical. This implies that simple sedimentation

experiments can be used to determine the pseudo-equilibrium concentrations.

The stability of montmorillonite colloids decreases with increasing ionic strength

and decreasing pH as expected. The pH and ionic strength were proved to be

critical parameters influencing the effect of temperature. At intermediate pH, the

effect of increasing temperature stabilized the suspensions with ionic strength

lower than 0.0068 M. At pH≤4 increasing temperature dramatically reduced the

stability of montmorillonite suspensions independently of ionic strength. At

pH≥10 increasing temperature stabilized the suspensions for all ionic strengths

studied.

Montmorillonite suspensions were more sensitive to ionic strength changes

induced by increasing CaCl2 than by NaCl. The CCC value obtained for Ca-

montmorilonite in a solution containing CaCl2 was two orders of magnitude lower

than that for Na-montmorillonite in a solution containing NaCl.

Latex and montmorillonite colloids were shown to sorb to minerals under

unfavourable conditions. This process appears to be governed by the electrostatic

forces between minerals and colloids as reflected by mineral pzc and colloid zeta

potential. Consequently, sorption is affected by water chemistry in a similar

fashion to colloid-colloid interactions.

Colloids break through the columns packed with different fracture filling minerals

under advective regime. Colloid sorption to minerals contributes to colloid

retention during transport reflecting pzc trends of the minerals. In the present

experiments, Na-montmorillonite retention is reversible while for latex colloids

retention is irreversible in the experimental time frame.

The DLVO theory provided a qualitative explanation for effects of the different

parameters on colloid generation, stability of the suspensions and sorption.

54

6. Acknowledgements

Many people have contributed to this thesis in many different ways. Therefore I would

like to thank:

- my two supervisors Prof. Mats Jonsson and Dr. Susanna Wold for all I learned from you

these years not only during our discussions. Your interest in science increased my

motivation.

- the Swedish Nuclear Fuel and Waste Management Co., SKB, for funding this work and

in particular, Dr. Ignasi Puigdomenech and Dr. Kastriot Spahiu for all the support,

discussions and good advices.

- Dr. Claude Degueldre for giving me the opportunity to visit the Paul Scherrer Institute

(PSI) and learn more about colloids, and to all the people that I meet there.

- to all of you that have contributed with diverse practical help: everyone at Nuclear

Chemistry (KTH) and in particular Michael Holmboe, Anders Puranen and Kjell

Svärdström for support in the lab; Dr. Alan Snedden and Dr. Andreas Fischer at Inorganic

Chemistry (KTH) for assistance with SEM and XRD analysis; Ola Karnland at Clay

Technology for providing montmorillonite samples; Oskar Sigurdsson at Äspö hardrock

laboratory, Sweden for assistance during mineral sample collection; Jan Olov Nyström

and Henrik Skogby at the Royal Museum of Natural History in Stockholm for providing

biotite samples; Luc van Loon, Martin Glaus and Sabina Frick at PSI for help with

generation experiments.

- I am also grateful to Prof. Emeritus Trigve Eriksen, Prof. Ivars Neretnieks, Michael

Holmboe, Dr. Luc van Loon and Martin Glaus for illuminating discussions.

- Thanks to all the former and present workmates from the Nuclear Chemistry and

Inorganic Chemistry divisions at KTH for creating a scientific and enjoyable atmosphere

at work.

- My family and friends here and there, thank you for the faith, the good advice and the

great interest that you have always shown in my work.

- Special thanks to Victor for your facility to listen, understand and see the positive in the

difficulties.

55

7. Bibliography

[1] Atteia, O., Kozel, R., Particle size distributions in waters from a karstic aquifer: from

particles to colloids. J. of Hydrology 201 (1997) 102-119.

[2] Dearlove, J.P.L., Longworth, G., Ivanovich, M., Kim, J.I., Delakowitz, B., Zeh, P., A

study of groundwatercolloids and their geochemical interactions with natural

radionuclides in Gorleben aquifer systems. Radiochimica Acta 52/53 (1991) 83-89.

[3] Gschwend, P.M., Backhus, D.A., MacFarlane, .I.K., Page, A.L., Mobilization of

colloids in groundwater due to infiltration of water at a coal ash disposal site. J. Contam.

Hydrol. 6 (1990) 307-320.

[4] Degueldre, C., Baeyens, B., Goerlich, W., Riga, J., Verbist, J., Stadelmann, P.,

Colloids in water from a subsurface fracture in granitic rock, Grimsel Test Site,

Switzerland. Geochem. Cosmochim. Acta 53 (1989) 603 -610.

[5] Sposito, G., The surface chemistry of soils, Oxford University Press, New York, 1984.

[6] McCarthy, J.F., Zachara, J.M., Subsurface transport of contaminants: Mobile colloids

in the subsurface environment may alter the transport of contaminants. Environmental

Science and Technology, 23 (1989) 496-502.

[7] Kersting, A., Efurd, D., Finnegan, D., Rokop, D., Smith, D., Thompson, J., Migration

of plutonium in ground water at the Nevada Test Site. Nature, 397 (1999) 56-59.

[8] www.andra.fr

[9] www.nirond.be

[10] http://www.posiva.fi/en

[11] www.skb.se

[12] Final storage of spent nuclear fuel - KBS-3, I General; II Geology; III Barriers; IV

Safety; - Summary. Swedish Nuclear Fuel Supply Co/Division KBS (SKBF/KBS) 1983

[13] SR 97-post-clossure safety, SKB TR-99-06, Main Report, Vol. I., II. SKB, 1999.

[14] Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first

evaluation Main Report of the SR-Can project, SKB Technical Report, TR-06-09, 2006.

[15] Segall, R.L., Smart, R.St.C., Turner, P.S., Surface and Near- Surface Chemistry of

Oxide Materials. Materials Science Monographs 47, edited by J. Nowotny and L.-C.

Dufour, Elsevier Science Publishers B. V., Amsterdam, 1988.

56

[16] King F., Corrosion of copper in alkaline chloride environments. SKB Technical

Report, TR-02-25, 2002.

[17] Börgesson, L., Hernelind, J., Earthquake induced rock shear through a deposition

hole. Influence of shear plane inclination and location as well as buffer properties on the

damage caused to the canister. SKB Technical Report, TR-06-43, 2006.

[18] Karnland, O., Olsson, S., Nilsson, U., Sellin, P., Mineralogy and sealing properties

of various bentonites and smectite-rich clay materials. SKB Technical Report, TR-06-30,

2006.

[19] Vejmelka, P., Fanghaenel, Th., Kienzler, B., Korthaus, E., Römer, J., Schüssler, W.,

Artinger, R., Sorption and migration of radionuclides in granite (HRL Äspö, Sweden).

FZKA 6488, 2000.

[20] Andersson, J., Ström, A., Svemar, C., Almén, K-E., Ericsson, L.O., What

requirements does the KBS-3 repository make on the host rock? Geoscientific suitability

indicators and criteria for siting and site evaluation. SKB Technical Report, TR-00-12,

2000.

[21] Bergelin, A., Lindquist, A., Nilsson, K., Nilsson, A.-C., Forsmark site investigation.

Hydrochemical characterization in borehole KFM10A. Results from three investigated

borehole sections: 298.0–305.1 m, 436.9–437.9 m, 478.0–487.5 m. SKB Technical report

P-07-42 2007.

[22] Wass, E., Groundwater flow measurements in permanently installed boreholes. Test

campaign no. 4, 2008. Forsmark site investigation. SKB Technical report P-09-30, 2009.

[23] Auqué, L. F., Gimeno, M.J., Gómez, J.B., Puigdomenech, I., Smellie, J., Tullborg,

E-L., Groundwater chemistry around a repository for spent nuclear fuel over a glacial

cycle. Evaluation for SR-Can. SKB Technical Report, TR-06-31, 2006.

[24] Boulton, G.S., Caban, P.E., van Gijssel, K., Groundwater flow beneath ice sheets:

Part I — Large scale patterns. Quaternary Science Reviews, 14 (1995) 545-562.

[25] Neretnieks, I., Flow and transport through a damaged buffer- exploration of the

impact of a cemented and an eroded buffer. SKB Technical Report, TR-06-33, 2006.

[26] Neretnieks, I., Liu, J., Physical and chemical stability of the bentonite buffer, R-06-

103, SKB, 2006.

[27] Karnland, O., Sandén, T., Johannesson, L.-E., Eriksen, T. E., Jansson, M., Wold, S.,

Pedersen, K., Motamedi, M., Rosborg, B., Long term test of buffer material. Final Report

on the pilot parcels. SKB Technical Report, TR-00-22, 2000.

57

[28] Puigdomenech, I., Hydrochemical stability of groundwaters surrounding a spent

nuclear fuel repository in a 100000 year perspective, SKB Technical Report, TR-01-28,

2001.

[29] Neretnieks, I., Diffusion in the rock matrix: an important factor in radionuclide

retardation?. J. Geophys. Res. 85 B8, (1980) 4379–4397.

[30] Vilks, P., Bachinski, D.B., Colloid and suspended particle migration experiments in

a granitic fracture. Fourth International conference on the chemistry and migration of

actinides and fission products in the geosphere. Charleston, SC, USA, proceedings.

Radiochimica acta, special issue, 66/67 (1993) 229-234.

[31] Norrish, K., The swelling of montmorillonite. Discuss. Faraday Soc. 18 (1954) 120-

134.

[32] Hunter, R. J., Introduction to modern colloid science. Oxford Science Publications,

1993.

[33] Shchukin, E.D., Pertsov, A.V., Amelina E.A., Zelenev, A.S., Colloid and Surface

Chemistry. Studies in Interface Science, vol. 12, Elsevier, 2001, p.214.

[34] Shaw, D.J., Introduction to colloid and surface chemistry. Butterworths & Co

Publishers, 1970.

[35] Ran, Y., Fu, J.M., Sheng, G.Y., Beckett R., and Hart, B.T., Fractionation and

composition of colloidal and suspended particulate materials in rivers. Chemosphere,

41 (2000) 33-43.

[36] Plaschke, M., Schäfer, T., Brundschuh, T., Ngo Manh, T., Knopp, R., Geckeis H.,

.Kim, J.I., Size Characterization of Bentonite Colloids by Different Methods. Anal.

Chem., 73 (2001) 4338-4347.

[37] Cadene, A., Durand-Vidal, S., Turq, P., Brendle, J., Study of individual Na-

montmorillonite particles size, morphology, and apparent charge. J. Coll. Interface Sci.,

285 (2005) 719-730.

[38] McCarthy, J.F., Degueldre, C., Sampling and Characterization of Colloids and

Particles in Groundwater for Studying Their Role in Contaminant Transport.

Environmental Particles, Vol.2, J. Buffle and H. P. van Leeuwen (Eds.), Lewis Publisers,

Boca Raton, FL, 1993 p. 247-315.

[39] Ryan, J.N., Elimelech, M., Colloid mobilization and transport in groundwater.

Colloids Surfaces A: Physicochemical Eng. Aspects 107 (1996) 1-56.

58

[40] Karnland, O., Olsson, S., Nilsson, U., Sellin, P., Experimentally determined

swelling pressures and geochemical interactions of compacted Wyoming bentonite with

highly alkaline solutions. Physics and Chemistry of the Earth 32 (2007) 275–286.

[41] Scatchard, G., Equilibrium in solutions. Surface and Colloid Chemistry”, p.260-261,

Harvard University Press, 1976.

[42] Stechemesser, H., Dobiáš, B., Coagulation and Flocculation. 2nd

edition, Surfactant

Science Series vol. 126, CRC Press Taylor and Francis, 2005, p.81-88.

[43] J. Lyklema, Fundamentals of Interface and Colloid Science. Vol. IV, Elsevier

academic press, 2005.

[44] B.V. Derjaguin, L. Landau, Acta Physicochim URSS, 14 (1941) 633.

[45] Reerink, H., Overbeek, J.Th.G., Theory of the stability of lyophobic colloids.

Elsevier, Amsterdam 1948.

[46] Hiemenz, P.C., Principles of colloid and surface chemistry. 2nd

ed. Marcel Dekker,

inc. New York, 1986.

[47] Gillham, R.W., Cherry, J.A., Predictability of solute transport in diffusion controlled

hydrogeologic regimes, Proceedings of the Symposium on Low-Level Waste Disposal:

Facility Design, Construction and Operating Practices, 1982a.

[48] Gillham, R.W., Cherry, J.A., Contaminant migration in saturated unconsolidated

geologic deposits, Recent Trends in Hydrogeology, The Geological Society of America,

Special Paper 189, Boulder, Colorado, 31–62, 1982b.

[49] Davis, S.N., “Porosity and permeability in natural materials. Flow through Porous

Media, DeWiest, R. J. M. (ed.), 53-89. Academic Press, New York, NY, 1969.

[50] Darcy, H., Les Fontaines Publiques de la Ville de Dijon. Victor Dalmont, Paris,

1856.

[51] Hiscock, K.M., Hydrogeology: principles and practice Blackwell Pub., Malden, MA

2005.

[52] Fetter, C.W., Applied Hydrogeology. MacMillan, New York, NY, 1994.

[53] Stumm W., Morgan, J.J., Aquatic Chemistry: Chemical equilibria and rates in

natural waters. 3rd edn., Wiley, New York, 1996.

[54] Malmström, M., Banwart, S., Lewenhagen, J., Duro, L., Bruno J., The dissolution of

biotite and chlorite at 25°C in the near-neutral pH region Journal of Contaminant

Hydrology, 21 (1996) 201-213.

59

[55] Sverjensky, D.A., Zero-point-of-charge prediction of crystal chemistry and salvation

theory, Geochimica et Cosmochimica Acta 58 (1994), pp. 3123–3129.

[56] Sondi, I., Bian, J., Pravdi, V., Electrokinetics of Pure Clay Minerals Revisited. J.

Coll. Interface Sci., 178 (1996) 514-522.

[57] Rossé, P., Loizeau, J.L., Use of single particle counters for the determination of the

number and size distributions of colloids in natural surface waters. Colloids and Surfaces

A: Physicochem. Eng. Aspects 217 (2003) 109-120.

[58] Liquid sampler-50 operator´s manual, P/N M10205 Rev A, Particle measuring

systems, Inc. 2004

[59] Ledin, A., Karlsson, S., Düker, A., Allard, B., Applicability of photon correlation

spectroscopy for measurement of concentration and size distribution of colloids in

natural waters. Anal. Chim. Acta, 281 (1993) 421-428.

[60] Xu, R., Particle Characterization: Light Scattering Method, Hingham, MA, USA:

Kluwer Academic Publishers, 2000. p 223-269.

[61] Miller, W.M., Alexander, W.R., Chapman, N.A., McKinley, I.G., Smellie, J.A.T.,

Geological Diposal of Radioactive Wastes and Natural Analogues. Waste Management

Series, vol. 2, Pergamon, Amsterdam, 2000.

[62] Vold, R.D., Vold, M.J., Colloid and Interface Chemistry. Addison Wesley

Publishing Company, 1983

[63] Skoog, D.A., Leary, J.J., Principles of instrumental analysis, Saunders Golden

Sunburst Series, 1992.

[64]http://www.varianinc.com/image/vimage/docs/products/spectr/fluoro/brochure/1757.

pdf

[65] Choppin, G.R.R., Rydberg, J., Liljenzin, J.-O., Radiochemistry & nuclear chemistry

2ed., Elsevier Science & Technology Books, 1995.

[66] Missana, T., Alonso, U., Turrero, M.J., Generation and stability of bentonite

colloids at the bentonite/granite interface of a deep geological radioactive waste

repository. J. of Contam. Hydrology, 61 (2003) 17-31.

[67] Bessho, K., Degueldre, C., Generation and sedimentation of colloidal bentonite

particles in water. App. Clay Sci., 43 (2009) 253-259.

[68] Pusch, R.. Experimental study of the effect of high porewater salinity on the physical

properties of a natural smectitic clay. SKB Technical Report, TR-01-07, (2001).

60

[69] Tang, L., Sparks, D.L., Cation-exchange kinetics on montmorillonite using

pressure-jump relaxation. Soil Sci. Soc. Am. J., 57 (1993) 42- 46.

,

[70] Frey E. and Lagaly, G., Selective coagulation in mixed colloidal suspensions. J. Coll

Interface Sci., 70 (1978) 1.

[71] Tombácz, E., Balázs, J., Lakatos J., and Szántó, F., Influence of the exchangeable

cations on stability and rheological properties of montmorillonite suspensions. Colloid

Polym. Sci., 267 (1989) 1016-1025

[72] Chheda, P., Grasso D., and van Oss, C.J., Impact of ozone on stability of

montmorillonite suspensions. J. Coll. Interface Sci., 153 1 (1992) 226-236.

[73] Lagaly G., Ziesmer, S., Colloid chemistry of clay minerals: the coagulation of

montmorillonite dispersions. Advances Coll. Interface Sci., 100-102 (2003) 105-128.

[74] SKB, 2004a. Interim initial state report for the safety assessment SR-Can. SKB

Report R-04-35, 2004.

[75] Homola, A., James, R.O., Preparation and characterization of amphoteric

polystyrene lattices. J. Coll. Interface Sci., 59 (1977) 123-34.

[76] Avena, M.J., De Pauli, C.P., Proton Adsorption and Electrokinetics of an

Argentinean Montmorillonite. J. Coll. Interface Sci., 202 (1998) 195-204.

[77] Kraepiel, A.M.L., Keller, K., Morel, F.M.M., A Model for Metal Adsorption on

Montmorillonite., J. Coll. Interface Sci., 210 (1999) 43-54.

[78] Tombácz. E., Ábrahám, I., Gilde, M., Szántó, F., The pH-dependent colloidal

stability of aqueous montmorillonite suspensions. Coll. and Surfaces, 49 (1990) 71-80.

[79] Compton, R.G., Pritchard, K.L., The Dissolution of Calcite at pH > 7: Kinetics and

Mechanism, Phil. Trans. R. Soc. Lond. A 330 (1990) 47-70.

[80] Alonso, U., Missana, T., Patelli, A., Ceccato, D., Albarran, N., García-Gutiérrez, M.

Lopez-Torrubia, T., Regato, V., Quantification of Au nanoparticle retention on a

heterogeneous rock surface. Colloids Surf., A. 347: 1-3, (2009) 230-238.

[81] Taboada-Serrano, P., Chin, C.-J., Yiacoumi, S., Tsouris, C., Modeling aggregation

of colloidal particles. Current Opinion Coll. Interface Sci., 10 (2005) 123-132.

[82] Filby, A., Plaschke, M., Geckeis, H., Fanghänel, Th.,. Interaction of latex colloids

with mineral surfaces and Grimsel granodiorite. J.Contam. Hydrol., 102 (2008) 273-284.